100% found this document useful (9 votes)
81 views84 pages

Immediate Download Building Enterprise Taxonomies 2nd Edition Darin L. Stewart Ebooks 2024

ebook

Uploaded by

leincakran
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (9 votes)
81 views84 pages

Immediate Download Building Enterprise Taxonomies 2nd Edition Darin L. Stewart Ebooks 2024

ebook

Uploaded by

leincakran
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 84

Full download ebook at ebookname.

com

Building Enterprise Taxonomies 2nd Edition Darin


L. Stewart

https://ebookname.com/product/building-enterprise-
taxonomies-2nd-edition-darin-l-stewart/

OR CLICK BUTTON

DOWLOAD NOW

Download more ebook from https://ebookname.com


More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Web Based Enterprise Energy and Building Automation


Systems 1st Edition Barney L. Capehart

https://ebookname.com/product/web-based-enterprise-energy-and-
building-automation-systems-1st-edition-barney-l-capehart/

The Forgotten Founders 1st Edition Stewart L. Udall

https://ebookname.com/product/the-forgotten-founders-1st-edition-
stewart-l-udall/

Enterprise Integration Patterns Designing Building and


Deploying Messaging Solutions 1st Edition Hohpe

https://ebookname.com/product/enterprise-integration-patterns-
designing-building-and-deploying-messaging-solutions-1st-edition-
hohpe/

Ernst Krenek The Man and His Music John L. Stewart

https://ebookname.com/product/ernst-krenek-the-man-and-his-music-
john-l-stewart/
Highway engineering handbook building and
rehabilitating the infrastructure 2nd ed Edition R L
Brockenbrough

https://ebookname.com/product/highway-engineering-handbook-
building-and-rehabilitating-the-infrastructure-2nd-ed-edition-r-
l-brockenbrough/

Essentials of Neuroimaging for Clinical Practice 1st


Edition Darin D. Dougherty

https://ebookname.com/product/essentials-of-neuroimaging-for-
clinical-practice-1st-edition-darin-d-dougherty/

Human Resource Management Linking Strategy to Practice


Second Edition Greg L. Stewart

https://ebookname.com/product/human-resource-management-linking-
strategy-to-practice-second-edition-greg-l-stewart/

Geometric puzzle design 2nd Edition Stewart T Coffin

https://ebookname.com/product/geometric-puzzle-design-2nd-
edition-stewart-t-coffin/

Enterprise Planning and Development small business and


enterprise start up survival and growth 2nd Edition
David Butler

https://ebookname.com/product/enterprise-planning-and-
development-small-business-and-enterprise-start-up-survival-and-
growth-2nd-edition-david-butler/
Buil
E .

• Darin
I
L. Stewart
Building
Enterprise
Taxonomies
A Controlled Vocabulary Primer

SECOND EDTTTON

Darin L. Stewart, Ph.D.

(@
Mokita Press
To Laura,
who taught me the importance of details.
Copyright © 2011 by Darin L. Stewart

Published by Makita Press.

All rights reserved. Printed in the United States of


America. No part of this book may be reproduced in
any manner whatsoever without written perm1ss1on
except in the case of brief quotations embodied in
critical articles and reviews. For information contact
[email protected]

SECOND EDITION
Contents

1. Findability 1
Infoglut 4

The Problem with Search 6


Tclepor ting and Orienteerin 12

2. Metadata 23
T ypes of Metadata 28
Descriptive Metadata 28
Administrative Metadata 30
Structural Metadata 31
Metadata Schemas 32
Where Do I Put It? 36
Where Does It Come From? 39
Metadata and Autho rity Control 41

3. Taxonomy 45
Linnaean Taxonomy 48
Controlled Vocabulat:ies 51
Faceted Classification 59

4. Preparations 67
The Taxonomy Development Cycle 70
Research 72
Performing a Content Audit 76
Creating a Governance Document 83

5. Terms 89
Internal T erm Sources 90
I ntranets and Websites 92
External Term Sources 95

E xisting Taxonomies 98
Refining Terms 99

Basic Hygiene 99

Compound and Precoordinated Terms 107

D isambiguation 110

6. Structure 115
Card Sorting 116

Categories and Facets 122

7. Interoperability 135
Basic XML Concepts 137

Representing Hierarchy 140

l ;ear of Baggage Handling 144

XSLT 146

Zthes 152

8. Ontology_ 159
What Is An Ontology? 160

Class Hierarchy, Slots and Facets 162


Resource D escription Framework 168
RDF/ XNU~ 176
RDF Schema 180
Web O ntology Language (OWL) 181

9. Folksonomy 185
Tagging 186
Folksonomy 191
Tag Clouds 194
Pace Layerin 202

Glossary 205
Notes 219
Index 227
1
Findability
"But the plans were on display... "
"On display? I had to go down to the cellar to find them."
'That's the display department."
"With a flashlight. "
"Ah, well, the lights had probably gone."
"So had the stairs."
"But look, you found the notice didn't you?"
"Yes," said Arthur, "yes I did. It was on display in the
bottom of a locked filing cabinet stuck in a disused lavatory
with a sign on the door saying 'Beware of the Leopard."'

From The Hitchhikers Guide to the Galaxy

rinding good information is hard, m uch harder than it should be.


Your first en counter with a new website often feels like en tering a
strange land with its own language, laws, customs and culture. You
have business to conduct th ere, but must do so witho ut the benefit
of an interpreter or guide. As you b egin to explore the homepage,
you m ust quickly orient yoursel f to its wugue approach to navigation,
in terpret bizarre labels and menus, guess at search terms and wade
through propagan da in search of useful informatio n. And these are
just the public pages.

T hings get much more dangerous if you venture out o f the tourist
areas and onto an i.ntranet or, heaven help you, a file system. Once
you enter the realm of the enterprise i.n fo nnation system , all bets are
o ff. The seemingly unified front o f the corporate website dissolves
into a collection of fiefdo ms, each with its own local dialect and
2 Chapter One

jealously guarded borders passable o nly with the right permissions


and passwords. There also seems to be a civil war underway.

D espite our best efforts, most websites, portals, intrancts and file
systems are hostile environments for information seekers. We hire
consultants, hold focus groups and conduct usability studies to
w1derstand our users' needs. We build site maps, add search boxes,
and tag our content, and users still get lost. According to surveys
co nducted by Gartner, IDC and o thers, knowledge wo rkers spend
fro m thirty to as much as for ty percent o f their work day searching
for information and yet only find what they need less than half the
1
time. This means we spend more time looking for documents than
actuaUy reading them. This situatio n is not just embarrassing, it's
expensive.

A third of a scmor knowledge worker's time, the time they spend


chasing information , works out to be ro ughly S26,000 a year i.n salary
and benefits on average. When those searches arc successful, this is a
legitimate cost of doing business. When they fail , that fruitless search
time is a drain on resources. Y ct as expensive as tlu s may seem,
search time is a mi.nor compo nent of the cost of luddcn in formatio n.
Even ilic tens o f iliousands of dollars spent o n redesigning and
maintaining an improved website is trivial if it gets users to tl1e
content iliey need. The true cost comes when users ilirow up their
hands and abandon ilieir search. Studies have suggested that this
happens after abo ut twelve minutes at tl,c outside.

This phenomenon is not restricted to co mplex searches and o bscure


facts. Inforn1atio n as mundane as tl,e contact i.n formatio n for the
director of human resources cannot be located by employees o n their
own Intranet fifty-seven percent of the time. Those intrepid few who
can find ilic information usually must troU tl1ro ugh multiple Web
pages and documents loo ki.ng for an org chart (which is probably out
of date) that nu ght have tl,e directo r's name. They must then look up
the director in an employee directory located elsewhere o n the
Intranet hoping they spelled ilie nam e right. O ne study found tlus to
be the case in five o ut o f si.-..:: corporate Intraoets.2
Findability 3

When people can't fi11d what they need, they don't just give up. They
go elsewhere. Wh en a consumer doesn't find the right product, they
go to a competitor, which in aggregate costs your company half of its
po tential sales. Tf they arc already a customer, they pick up the pho ne.
This costs you an average of seventeen dollars for each call that yom
self-service website was supposed to eliminatc. 3 When an employee
can't find what they need, they go to a co-worker, doubling costs
while halving productivity and often yielding no better results. Tn a
2002 research note, Rcgi11a Casonato and Kathy H arris of G artner
estim ated that ao employee will get fifty to seventy-five percen t o f
the mformation they need directly from other people, effectively
erasmg tl1e benefits o f a corporate Tntranet. 4

When a knowledge worker reaches this dead-end, they have little


choice but to set about creating tl1e in fo rmation they need &om
scratch. T his may be as simple as running a report and stitching a few
documents together, but more o ften it involves considerable
research, an additio nal in formatio n chase and consultation with
multiple colleagues. Unfortuna tely, all this effort is no t being
expended to create information, but to recreate it. As much as ninety
percent of the time spent creating information for a specific need is
actually recreating information that already exists but could no t be
5
located. According to Kit Sims Taylor, this is because it is simply too
hard to find what you need.

At present it is easier to write that contract clause,


exam question, insurance policy clause, etc., ourselves
than to find something close enough to what we want
from elsewhere ... . While most of us do not like to
admit that much of our creative work involves
reinventing the wheel, an honest assessment of our
work would indicate that we do far more 'recreating'
than creating. 6

Taylor has found that in addition to the amoun t of time spent


looking for information, an additio nal thirty percent is spen t
reinventing the wheel. When you account for communication and
collaboration overhead, only ten percent o f our time, effort and
4 Chapter One

energies is actually spent in the creation o f new knowledge and


information. In a separate study, IDC found that th.is "knowledge
work deficit" costs Fortune 500 companies over twelve billion dollars
annually.7

These arc just the purely quantifiable costs. Consider the impact poor
findability has on decisio n making when there simply isn't time to re-
research and recreate the needed intelligence. Critical decisions may
be delayed because the information we can find, if any, is either
incomplete or conllicting. Worse, bad decisio ns may be enacted when
they wouldn't have even been considered had a fuller, more accurate
picture been available. ln this age of compliance, the ability to locate
and produce in form ation on demand can mean the difference
between passing an audit and dissolving the company.

lnfoglut

So how did we get into this mess? We have spent literally trillions o f
dollars on information technology, and yet o ur access to in formation
seems to get worse in direct proportion to the amount o f money and
effort expended to improve it. Some pundits point to the sheer
volume of in formatio n with which we arc inundated and resign
themselves to this inevitable consequence of life in the information
age. As Britton H adden of n NfE magazine put it:

Everyday living is too fast, too busy, too complicated.


More than at any other time in history, it's important to
have good information on just about every aspect of
life. And there is more information available than ever
before. Too much in fact. There is simply no time for
people to gather and absorb the information they need.

Hadden made this o bservation in 1929, shortly before founding the


magazine. Infoglut is not a new problem, but until recently it was at
least somewhat manageable. Today we are discovering that the.: only
Findability 5

thing worse and more dangerous than trying to run an organization


with too little information is trymg to manage o ne with too much.
Everyone understands intuitively that infoglut is a problem, bu t few
have a clear sense o f how m uch of a problem it really is. Experts
have lo ng proclaimed the dangers of information overload. While
hyperbole is the lifeblood of consultants, in this case they seem to be
right o n th.e money.

Each year the world produces roughly five exabytes (1018 /ijleJ) of
new in formation. T o put that in more familiar terms, if the seventeen
million books in the Library of Congress were fully digitized, five
exabytes would be the eg uivalent o f 37,000 new libraries each year.
While thjs is staggering in and of itself, consider that in 1999 it is
estimated that only two exabytes of new information was created,
meaning that the rate of in formation growth is accelerating by 30% a
year. 92% of that information is stored on digital meilia and 40% is
generated by the United States alone. We create 1,397 terabytcs of
o ffice documents each year. Each day we send thirty-one billi.on
emails.8 It is no wonder that we are, as J o hn Naisbitt famously put it,
"drowning in information, but starved for knowledge."

The deluge bas not caught us by surprise. O n the contrary, we have


attacked it with a vengeance, pouring billions into data warehouses,
CR.N(, EJUJ, business intelligence and other data management and
reporting systems. These efforts and investments have bought us
great insight into our str11ctured con/en!: that highly organized
informatio n structured according to a well defined schema or
framework. These are the records found in relational databases and
tl1at slot so ruccly into spreadsheets and reports. 'l'he information
contained in these records can easily be located, manipulated and
retrieved by means of standard guery languages such as SQL.

U nfortunately, this type of domesticated data makes up only fifteen


percent of the total in forma tio n with which we must copc.'1 T he
remaining eighty-five percent is made up of Web pages, emails,
memos, PowerPoint presentations, invoices, product literature,
procedure manuals, take-out menus and anything else that doesn't fit
neatly into a row in a database. The commo n factor among all of
6 Cb ap terOne

these different forms of 1111struct11red co11/e11! is that they arc all designed
for human consumption rather than machine processing. As a result,
all of the tried and true methods of data management we have
worked so hard to master fail miserably when asked to bring a
company picnic an nouncement to heel. So while quarterly sales
forecasts across four continents may be readily available, knowing
whether you are supposed to bring a salad or a desser t may be out o f
reach.

The Problem with Search


This aspect of the information o nslaught bas in fact taken us by
surprise. Many o f us arc still in denial. A fter all, with fully indexed,
electronic in formation sources, full-text searching should allow us to
specify all the terms and subjects in which we are interested and have
the information retrieved and delivered to our desktop. As any user
of Google, A9, or countless other search and retrieval e ngines has
learned tlirough painful experience, things rarely work out that neatly.

Rather than receiving a nice, neat set of t,'l.rgeted documents, search


engines generally present us with long lists o f Web pages that merely
contain the words o n which we searched. Whether or not th ose
words a.re used in the manner and context we intended (did you
mean Mercury the planet, tl1e car, ilie Roman God or the element?)
isn't pa.rt of the equation. We a.re left to sort thro ugh page after page
of links looking for something that might be relevant.

Part of th.is problem is self-inflicted. People just don't write good


queries. O ne third of the time, search engine users only specify a
10
single word as tl1e.i.r query and on average use only two or thrcc.
This is what leads to so many irrelevant documents being returned.
We do n't give enough context to o ur subject to elimi nate documents
th at arc no t o f interest. lf you query just o n the term "Washington"
you will receive links to in formatio n o n th e state, the president, the
capital, a type o f apple, a movie star, a university and so forth. In all,
Google returns 1,180,000,000 "hits." If you add the term "Denzel"
Findability 7

the number of links drops to 3,520,000, and we are reasonably


focused on the actor. If we add the phrase "Academy Awa.rd" we
finally get to 107,000 docwnents reasonably focused on the actor's
accolades. So the more specific and verbose we are with ou.r queries
the more relevant the results.

But what happens if you use the J\ cademy Awa.rd's comm on


nickname "Oscar" in yom query? The number of hits jumps to
593,000. This is the risk o f getting too specific with search terms. By
using the proper name of the award rather than its popular name, we
may have missed 486,000 potentially relevant documents. Guessing
the wrong search term can have a dramatic impact o n what you do
and don't frnd.

I nformation scientists have lo ng been aware that there a.re tradcoffs


between depth and coverage whenever a search is conducted. The
broader the sear ch is, the more documents that a.re retrieved,
including those that a.re not r elevant to the actual in formatio n need.
Conversely, the deeper or narrower the search, the more likely
retrieved documents a.re to be relevant. The cost, of course, is that it
is also more likely that documents of interest will be missed in the
search. T he difficulty arises from the fine balance of preciJion and
recall.

Precision is usually described as a ratio: the number o f relevant


documents retrieved divided by the total number of documents
retrieved. I n other words, what percentage of the to tal number of
do cwnents retrieved arc actually related to the topic being
investigated? For example, a Google search on the terms "precision"
and "recall" returns approximately 970,000 documents. T he fust few
documents in the list do indeed prove to be related to measures of
search performance. Howev er, a few links in to the list a news item
appears: "Vermont Precision Woodworks A nno unce Recall of
Cribs." From the search engine's perspective, this is a perfectly valid
docw11ent. Tt contains bo th of the search tcrn1.s in its title. In fact,
o ne search term appears in the title of the website itself, www.recall-
wa.rnings.com, thus causing it to receive a high relevancy ranking.
O ut of 970,000 documents, it is safe to assume that many, if not
8 Cbapter One

most, of the retrieved documents will have this level of relevancy to


our 9uery. This indicates low precision, but high recall.

Recall is also a ratio and is defined as the number of relevan t


documents retrievedthe total number o f relevant documents in the
collection being searched. The example above probably has a high
recall due to the large number o f documents returned.

Relevant Relev,ml
Documents Retrieved Documents Retrieved
Total
= PRECISION Total Relevant
= RECALL

Documents Retrieved Documents in Collection

recaU

These two measures are inversely related: as recall increases, precision


decreases. A balance must be fo und between the two, retrieving
enough docw11ents to get an individual the in formation they need
without returning so many that wading through irrelevant
information becomes burdensom e. This balance is the heart of
in formation retrieval, bu t it is difficult to measure precisio n and recall
precisely. 'Ibis is because we rarely know what is contained in the
collectio n we arc searching, in this case the Internet itself, and also
because the notion of relevance is very subjective. At best we can
estim ate recall and precisio n based o n feedback fro m users o f the
search engine in 9ucstion and make adjustments as appropriate.

Taking our Google search o n "precision" and "recall" as a test case, it


m ay seem that the problem isn't so bad. J\ fter all, the first several
documents in the list were o n the exact to pic we were seeking: search
performance measures. We can just disregard the other 3.5 million
documents o ffered. We got what we needed fro m the top ten or
l:'(vcnty.
Findability 9

This ability to rank pertinent documents near the top of a result set is
what has made Google the clear winner of the search engine wars.
T heir PageRanJ<: algorithm is a key ingredient in the Google secret
sauce. 11 Rather than just counting how many ti.mes a certain word
occurs in a document or where it occurs, Google also looks at who
links to that document. If a lot of pages reference a particular
website, chances are that it is a pretty im portant source of
information on the topic at hand. If the pages linking in are
themselves important, then that likelihood increases and the
document's relevancy rank improves accordingly. T his variation o n
"citation analysis," which is traditionally used to determine the
importance of scholarly publications, has radically changed Internet
search for the better. Google even offers a free tool that l can add to
my website to search my own content with just a few lines of code.

So, problem solved? Not quite. There arc several caveats to applying
a G oogle-like tool to your fi.ndability challenges. First, Google free
site search is really only searching a su bset of the entire Google index,
that part representing just your website. As a result, only those Web
pages that are open and available to the public will be included in a
search. Anything o n the l ntranet is invisible to the Google spiders,
th e programs that find and index Web pages and build up the search
index. Even those pages and documents that are open to the Internet
at large may be missed. Indexing programs only go so deep when
looking over a website. If your content is more than a link or two
away fro m the main page, it will probably be missed. Any new
content you add will likewise be invisible until the next time an
indexing spider happens by- a process completely outside of your
control. As Google explains:

There are a number of reasons a page might not appear


in t he results of your Google free site search. It could
be that Google hasn't crawled t hat particular page yet.
Google refreshes i ts index frequently, but some pages
are inevi tably missed. Or, the page may have
Javascri pt, frames, or store information in a database.
Pages like these are difficult or impossible for the
12
Google crawler to visit and i ndex.
Chapter 0 11e

Finally, Google's greatest strength, the PagcRank algorithm, is also its


greatest weakness when applied to a single website. l t is unlikely that
CNN.com or eBay will reference your org chart. In fact, very few
websites outside of your organization will link to your internal
documents. Yet the rankings applied to your documents are
determined in the context of rankings of the Internet as a whole. Th.is
effectively renders the relevancy judgments made on your content
meaningless when the search is restricted to your own sitc. 11

Aside from the arcane nature of indexing, the very act o f searching
can be a struggle in most organizations. Documents and content are
spread out across multiple locations and repositories. Policies may be
o n the Intranet, quarterly reports o n the file system, resumes in a
departmental directory and price lists on the company homepage.
Finding in formation is n o longer an exercise in finding a needle in a
haystack. First you must choose which haystacks to search, in what
order, and for how long. In most organizations, less than half of their
documents arc centrally inclexecl. 14 Th.is means that it is impossible to
look for information in aU potential locations with a single query or
even a single search tool. Th.is dispersal of information across an
organization leads to anoth er search challenge: choosin g the correct
query terms.

< !-- sitesearch Goog l e -- >


<F ORM rnethod =GET a ccion= " http: / ;..,...,. google . com/search " >
< input type=hidden n arne=ie value=UTF-B >
<i n p u t type=ludden narne=oe val ue=UT F-B >
< T ABLE £.9.£9.-!-..2£.= " #FFFFF F " ><~..i:.:><td>
< A HREF= " htt p : / /www. google . com/ " >
< IMG SRC= " ht tp : / /www . google.com/ l ogos/ Logo _ 40wht . gi f"
£2.uie r= " 0 " ALT= " Google " ></ A >
</ t d><td>
< INPUT TY PE=text n arne=q size=31 maxlength=255 v alue= "">
< INPUT type=submit n ame =btn<q VALUE="Google Search " >
< f ont size=-1>
<i n put type=hidden narne=dornains v alue="YOUR DOMAIN NAME " ><b.r_>
< input type= ra dio name=sit;es;ean:::J~ value= "" > WliJW <input type= r adio
!ls!l!!..~ i t~ ~ value="YOUR DOMAIN NAME" chec k e d > Y OUR DOMAIN NAM E
<b r ></font ></td></ t r></TABLE >
</ FORM>
~ Sit;eSea~ Google -->

Figure 2. Just cut, pas te and yo u've got search. Not quite.
Fir1dability 11

Most search engines create their


indexes by extracting terms
from the full text of
documents. As a result content
creators and authors become de
facto indexers and catalogers.
The wo rds they choose in
authoring their documents
beco me the search terms
available to their readers. This
becomes a problem if they
do n't speak the same language.
Th.is goes back to the Mercury
(planet, car, god) and actor
(Academy J\ ward or O scar) Fig ure 3. The idea l relationship
pro blem. between author and searcher.

U nless there is a company standard for terminology, and these are


rare, each area of an enterprise is going to have its own language. f\
cmtomer in one area may be a client in another and a patron somewhere
else. Th is lack of consistency in search and indexing terms has
proven to be the single greatest challenge to the effectiveness of
search and findability in general. 15

Ultimately, any search consists of, at rruru.mum, four hurdles that


must be cleared. First, the information seeker must be able to
articulate what they are lo oking for with the right syn tax for the
specific search tool being used. ext, they must guess what words an
autho r may have used to express the concept o f interest. T hen, with
the query in mind, they must figure o ut the most likely place to
search. Finally, they must sift through the results of their search,
separating the potentially relevant from the clearly irrelevant and
hope wh at they end up with is complete, representing all that is
available. Really, it's a wonder that we ever ftnd anything at all.
12 Chapter One

Teleporting and Orienteering


A keyword search is most often an attempt (usually several attempts,
actually) to go directly and instantaneously to the exact location o f
desired information. If we search the Web on the terms "Aladc:lin
Theater Box-Office" we h ope to land where we can purchase tickets
for concer ts at th.is small venue in Portland, Oregon, without having
to sift through irrelevant information. The academic commwlity has
labeled tllis sort of information seeking behavior teleporting. 16

Teleporting is o ne strategy for finc:ling information and can be


executed in vario us ways with a number of search tactics. In addition
to keyword search, an information seeker may attempt to tcleport by
specifying a specific URL, opening a certain email, or typing in a
directory path to a particular document. Perfect tclcporting (hitting
your target o n the first attempt) is a rare accomplishment; so rare in
fact that a game, "Google Whacking," has sprung up around the
challcnge. 17 Y ct despite tl1e difficulty in finding just the right
information witl1 search alone, most websites and information portals
seem designed to encourage the attempt as evidenced by the
ubiquitous search box.

A more realistic scenario is to teleport into the gen eral vicinity of the
informatio n you arc seeking, using search or some other tactic, an<l
then zero in on your target with a succession of small steps. T o buy
OLLr concert r-jckcts for a show at the AJadc:lin, for example, we might
teleport by typing in th e URL for tl,c theater: www.aladc:lin-
tl1catcr.com. We know we arc close, but still can't buy our tickets so
we may follow tl1c link to the "Upcoming Shows" page. Herc we find
the performer we arc loo king for listed \vitl1 a linl< to "show details"
so we click through to tl1at page. Finally we sec a banner for "Local
Ticket Outlet Information," which leads us to a link for the "Aladdin
Theater Online Ticketing Page" where we can order our tickets.

This strategy of locating information by continua lly narrowing o ur


search through incremental steps h as been dubbed orienteering
(though most people simply call it browsing) and has proven to be
Findability 13

the preferred approach to finding informatio n. Studies conducted at


the MIT Artificial Intelligence lab have found that in formation
seekers use keyword search less than forty percent of th e ti.me.
Surprisingly, this holds true even when searchers know exactly what
18
they arc looking for and even where to find it (see table 1).

Specific General Specific


Total
Information Information Doc ument

Orienteering 47 19 41 120

Teleporting 34 23 17 80
Total 81 42 58 200

Table I. Information need by search strategy ( 19 unknowns removed).

There are circumstances where keyword search yields nominally


better results than navigation. In one study, in formation seekers were
more successful at locating information on a well indexed medical
information site by using search rather than browsing. Interestingly,
those most successful at finding what they were looking for were
tl1ose individuals who turned to search only after browsing failed (see
figure 3). Even when individuals abandon browse oo a given
information hunt and succeed with search, they invariably return to
orienteering oo thei.r next task. 19

M.I.T. researchers have found several reasons why people prefer to


zero in on information rather than attempting to pounce o n it in a
single great leap. First, it can be difficult to dearly articulate exactly
what it is you arc seeking. This is the case even when trying to
retrieve familiar information and documents. Think o f the last ti.me
you were asked for directions to a familiar destination. Even though
you may be able to drive there without thinking, you may have a hard
ti.me giving step by step instructions on how to get to that same
location. Browsing reduces the cognitive demand on information
seekers by allowing them to follow familiar paths to the general area
of the information they are seeking, guickly and easily reducing the
size of the area they must explore. This also allows searchers to draw
14 Chapter One

Search versus browse success rates

100%
90% /
,,
80% ,. ,
-
Q)

I ll
0:::
V,
70%
60%
,

, , ,.
,
,., ,
, ,.
V, 50% , /
Q)
,'
40% ,. ,
/
0
0 /
:,
(f) 30%
/
,.,, '
20% ., , /
'
10%
0%
Browse Search Sea rch after
Browse fa ilure
Information Seeking Strategy

First choice of strategy

100% ,
/1I

....a.
(/) 80% -✓
/

/
.,,,,l ~

60% · / /
E

-
....
(I)

c:x:
40%
20%
,.,.

/
,,
/

0%
Brow se Search
Information Seeking Strategy

Figu re 4. Info rmation seeking behaviors.


Find:,bility 15

on a broad range o f "meta-in formatio n" about the target o f their


search.

For example, say you need to locate a company memo that was
circulated six months ago and has since disappeared into the bowels
of the company Intranet. E ven though you have no idea where to
find the memo itself, you recall seeing it referred to in a11 email from
a colleague. You may not know exactly where to find that email
either, bu t you likely will recall who it was from and roughly when
you received it, alo ng with some idea of the subject line and general
content. T his will allow you to find tl1c email that will in turn point
you toward the actual target o f your search-the company memo.
Even tho ugh you can't teleport even i_n to the general vicinity of the
memo, you can start from a known frame o f reference (the email)
and follo w clues along the way until you arrive at your goal.

The small steps of orienteering and the clues found along the way
also provide information seekers with a stro ng sense of locatio n
through out their search. The importance of tl1e "you arc here" factor
should not be underestimated. When users feel in control and that
tl1cy arc heading in the righ t direction and arc able to backtrack if
they take a wrong turn, they are less likely to abandon a search
prematurely. When people drop into the middle o f an information
space as a result o f a keyword search, they have no context and little
indicatio n o f how to proceed. T his sense of disorien tation can cause
both knowledge workers and potential customers to leave a website
as quickly as they arrive.

By contrast, navigating through an infomiation space allows the user


to become acclimated to the environment at their own pace, much
like easing into a hot bath rather than plunging into scalding watet:.
This process of guided explo ration also has the dual ben efit o f
building context for interpreting the target in formation on ce it is
found and allowing for serendipitous discoveries along the way. Most
importantly, informatio n seekers arc more likely to continue their
search i_f they are confident that they are on the right path and tl,at
their efforts will pay off.
16 Chapter One

t!.t:U.g use1t.com md . o rg Ask Tog

NN/g Nielsen Norman Group


Strategies to ennance-'tne"us"er experiehce - - - C _-

Home ~ services Publications ~ A bout NN/g


NN/g Home :• Services • T raining > Intranet usabili t y

Figure 5. Breadcrumb trails arc often used to give users a sense of control
over their exploration of a new information space.

It's interesting to note that the word browse derives from an


antiquated French term brost m eaning "young sh oot" and referring to
the way that animals feed on the young sh oots of trees and shrubs.
As animals seek for nourishment, they must balance tl,c n utrition to
be gained against tl1e energy expended obtaining it. T his behavior is
fundamentally the same for information seekers. Visitors to an
information space, whether it be a website, Intranet, database, file
system or what bave you, arc continually balancing cost and benefit:
"Will tl,c i.nformation I find here be worth tl,c time and effort it is
costing me to track it down?" As they browse a website, they will be
repeatedly assessing tlie likeW1ood of fmding what tl1cy need i.n tl1c
current enviro nment and determining when it's time to move o n to
more promising pastures.

This metaphor has become the basis of information foraging


theory, a model of information-seeking behavior developed by Peter
Pirolli and Stuart Card of the Xerox Palo Alto Research Ccnter.w
According to this model, we search for information across tl,c
Internet using essentially the same strategics hunter-gatherers use to
search for food across tl,c savannah. The nature o f the prey may be
new, but tl, c fundamental approach hasn't changed for millennia.
Botl, animals a nd humans attempt to maximize their "benefit per unit
cost. " When the benefit, in terms of likelihood of finding tl,e
necessary food or information witl, an acceptable investment o f time
and energy, falls below a certain tliresh old, the current website or
watering hole will be labeled sterile and the forager moves o n to a
more fertile patch. Steps can be taken to reduce the W<elihood of
users leaving o ur i_nformacion patches prematurely. One of tl,c most
Findabifity 17

effective strategies is to in crease the strength of the in formation scent


present in our systems.

The notion of information scent is central to information foraging.


The basic idea is that just like a game anim al, i.n fonnation leaves
behind spoor that can be detected and tracked.

Associated concepts "rub off' on one another, leaving


detectable traces, just as a watering hole frequented
by woolly mammoths will smell of woolly mammoths. A
hunter-gatherer seeking mammoths is likely to be
drawn to the watering hole, if only to look for spoor.
Information foragers do the same. Imagine you're
looking for texts about foraging theory. If [a search]
throws up a box containing the keyword "hunter-
gatherer", you're likely to select that box. It just smells
right. 21

Consider oux ticket purchasing example. When we fast arrive at the


theater's ho mepage, we sec labels such as "Artist of the Month" and
"Show Listings," which may even include the concert we arc seeking.
Even though we don't see that we can purchase tickets here, the page
smells like concert tickets so we continue our search by clicking on
"Upcoming Shows." Herc the scent gets stronger when we find the
right show along with a link to "Show Details," which finally gets us
to "Buy Tickets O nline. " Through out the process of browsing, th e
scent of concert tickets is strong and gets stronger the closer we get
to our goal. This continual positive feedback can keep information
seekers happy with the current infom1-ation patch and preven t them
from jumping to a competitor or colleague to meet their needs.

Strong information scent can be a double-edged sword if mishandled.


The most common pitfall occurs when a strong scent points toward
what should be the right answer but isn't. J akob Nielsen
demonstrated this phenomenon in a study of a health in formatio n for
teens wcbsite. 22 Users were asked to find out how much they could
weigh without being considered overweight. Most users quickly
18 Chapter One

gravitated toward an area of the site labeled "rood & Fitness." T h.is
clear, concise label had strong information scent fo.r the question at
hand. Featured pro minently within that area of the site was a lengthy
article entitled "What's the righ t weight for my height?" that was also
ranked highly by a search on the tenn "weight."

This would seem to be a bull's-eye except for the fact that the article
docs not contain the answer to the question. Because the information
scent leading to this article was so stro ng, users were convinced they
wer e looking in the right place. When th.c information wasn't there,
they naturally concluded that because it wasn't where it sho uld be, it
must not exist anywhere on tl1e site and abandoned ilieir search. T h.is
is an w1 fortunatc result since ilie answer was in fact available o n the
site. It was buried in an article titled "Body Mass Index (BMI)." T he
information scent of ili.is title for answering the target questio n is
almost non-existent. hrst, the title is a bit academic and maybe even
intimidating for the website's teenage audience. Worse, the title gives
no indication of the article's content which includes a straightforward

Thursday, Augu st 26
Pink Martini
Oregon Zoo
[ IIY IICIEISOILIM{ J Tlckot Price: $32.00 adv/ $32.00 d os
OoorJ nl Gntei
All A.Qes Event
ti -IPl"I. Lawn Entry @ SPl'I, Shol!f n t 7 PM
loullkkrt
...
O.tlrt l1fo,mtti1n

~ Does anvth 1no say summer 1n Porttand quite like a Prnk Maroni concert at the Oreoon Zoo? What
~-· better way t o hear sonos from their latest release - the lush, breezy "Splendor m the Grass" -
than on the zoo's lu sh, breezy concert lawn? Our hometown heroes are international stars, but
desprte a busy European tounno schedule, Pink Ma1t1nt will out away its oassports for two soeaal
performances at the zoo - their only Portland appearances this summer

• ,._ ' please note only GA tix are available at the Aladdin Uox office; reservation pack6ges ore
available at tickel11H1stcr.com.The concerts all start at 7 p,m. Your ticket will allow you into
the Zoo al 4 1,.111. ol tho day or the concert. For all concerts , the l nwn is close d al 4 p.m. for
• th e sound c.hcck. ond then ope ned ot 5 p.m. for concert tltkct· holders. • • • •
PiM!Mutini

Figure 6. A website with good informa tio n scent.


Findability 19

calculation of optimal weight using height, weight and age. In a


nutshell, the container o f the information was mislabeled.

The problem o f bad labels strikes at the heart o f findabilily. If


information seekers cannot recognize the content they are searching
for even when they find it, it may as well not exist. Even when an
information producer gives careful consid eration to labeling and
categorization, the result may have no meaning to information
consumers. J\ physician, wanting to be precise, may label a document
on treating a particular rcspiratoty condition with the terms
laryngotracheobronchitis, inspiratory stridor and dexamethasonc.
While this may be perfectly appropriate for other doctors, it is o f
little use to a mother searching the Web for information on how to
alleviate the wheezing cough of her daughter with croup.

Most information systems today are organized much like libraries


before Melvil Dewey created his decimal system for classification.
Patrons were left to wander stacks of untitled o.r oddly titled books
piled o n shelves according to some idiosyncratic organizational
scheme comprehended only by an arcane priesthood of local
librarians.

Overcoming this barrier to discovery is the role of controlled


vocabularies and taxonomies. By developing a structured collection
of terms and guidelines around how they arc to be applied,
information can be managed in a manner tl1at facilitates its discovery,
interpretation and use to the greatest extent possib le.

Beyond just finding information, the hierarchical nature of a


ta.xonom)1 can help educate an information seeker by guiding them
tluough a subject. The mother searching for information about her
daughter's illness will not only discover that dexamethasone is a
steroidal treatment for the condition, but that humidified air may also
alleviate her discomfort. Continuing tluough tl1c structure she will
discover additio nal treatments and potential complications. Finally,
she will learn that the proper name for "croup" is 1.11 fact
laryngotracheobronchitis, giving her a new term to search on and
expanding the potential information sources available to her.
20 Chapter One

The parent/ child relationships inherent in th e tree structure of a


taxonomy are powerful tools in guiding a seeker through what may
be an un familiar subject. By explicitly showing how terms and
concepts arc related, a searcher will discover associations that they
didn't: know existed. Most importantly, they can define and refine
their information need as they explore rather than having to precisely
articulate it up front wben they may not know exactly what it is they
are seeking.

O rgani7.ing informatio n according to a well defined structure, such as


a taxonomy, also provides stability to an information environment.
Information changes continually. D elphi Group has estimated that at
least ten percent of enterprise information changes mo nthly i.n an
average organization.23 Without some means o f governance, relevant
informatio n becomes a moving target. Today a search on
"taxono mies" may yield 1,900,000 matches. T o mo rrow o r next week
tlrnt same query could return 1,985,000 hits with completely different
ra nkings. That article I found last week that was so useful but that I
didn't bookmark could now be anywhere.

A taxonomy can act as a dynamic bookmark. As new documents and


in formatio n become available, they can be classified, labeled and
published in accordance with the taxono my without changing its
structure. When a knowledge worker needs to return to an area of
interest, he will still find it where he left it. The o nly difference will be
that tl1crc is now more information available there. In addition, the
new information will be in context witl, relationships an<l potential
avenues o f exploration clearly visible.

Managing terms and keywords can also enhance search by bridging


the vocabulary gap between information producer and consumer. A
search engine integrated witl, a ta.,xonomy would know that a search
on cro11p sho uld also look for laryngolracheobro11chitis and that in certain
contexts "Oscar" is an o ther way of saying "Academy Award." It can
also compensate fo r common spelling errors and variants (i.e., theatre
or theater) and syno nyms (fall or plunge or spill or tumble). T hese
expansions may seem trivial, but they can dramatically improve the
effectiveness and efficiency of search.
Findability 21

A sample hierarchy of respiratory illnesses

CROUP
(USE FOR laryngotracheobronchitis)
Symptoms
fever
wheezing
(USE FOR inspiratory strider)
swollen lymph glands
decreased appetite
Treatment
humidified air
fever reducer
acetaminophen
ibuprofen
steroid
dexamethasone
prelone
orapred
pulmicort
breathing treatment
acemic epinephrine
Complication
kidney inflammation
(USE FOR glomerulonephritis)
rheumatic fever

STREP THROAT
Symptoms
fever
swollen lymph glands
rash
Treatment
antibiotic
amoxicillin
erythromycin
Complication
rheumatic fever

RESPIRATORY SYNCYTIAL VIRUS


(USE RSV)
Symptoms
22 Chapter One

Controlled vocabularies, like taxonomy and its relatives, arc no t silver


bullets and will not magically cure all information management
problems, but they are a critical component of findability. If properly
constructed, applied and maintained, a ta,x ono m y can radically
increase the value of information by making it more available,
understandable and actionable. The remainder of this book will
demo nstrate how this can be achieved. Before we can delve into the
mysteries and wonders o f taxonomies, however, we must take a brief
detour into the world of metadata.
2
Metadata

If we fail to anticipate the unforeseen or expect the


unexpected in a universe of infinite possibilities, we may
find ourselves at the mercy of anyone or anything that
cannot be programmed, categorized or easily
referenced.

Fox Mulder, "The X-Files"

Art collecting is a tricky business. The value of a painting, sculpture


or even a rare book can vary wildly depending on the circumstances
of a purchase. Two similar works by Monet may go on the auction
block together; one sells for thousands, the other for millions. The
only substantive difference between the two is the existence of
provenance information. A clear record of a painting's histoiy, who
has owned it, when and where it has previously sold and for how
much is essential to deterrnio.i.ng whether or not it is a wise
investment. Without such information we have no context for our
decision. Is it overpriced or undervalued? Is it stolen? Is it a verified
Monet or just a suspected Monet? Even though it is the painting
itself that holds our interest, we need information about the painting
to gualify our interest. This same principle applies to less tangible
assets- namely in formation.
When we first locate new information we tend to be suspicious. Can
I trust these numbers? Is this the current version o f the document? Is
this image copyright cleared? This is especially true if the source of
24 Metad:1ta

that informatio n is no t familiar to us. Before we trust a document or


a Web page, we need to know a little more about it. Some of these
gucstions may be answered b y the search itself. When we lo ok for
information, we usually try to specify parameters to limit th e scope of
the search. Specifying the author of a document, the date o f its
p ublication, whether it is a report, invoice, form or memo will not
onJy enhance o ur chances of locating what we a.re looking for but can
pre-gualify the content as it is found. This kind of reference
information is generally not indicated explicitly in the content itself,
but rather is supplemen tary to it. I t is metadata.

The standard definition o f metadata is usually given as "data about


data." Th.is gets at the general idea, but is not gu.ite adequate. The
term "meta" co mes from the Greek root meaning something !hat jollmvs
anolher and lakes ii into acco1111t. Thus, metadata is gen erally developed
from associated source data and as a fun ction of the in formatio n it
describes. The G reek tem1 aJso means among, alongside, or 1vith, so it
follows that mctadata ca n take several complementary forms in
relationship to its paren t informatio n. rinally, if tl1c Latin derivatio n
is taken into account, meta can mean /ranscendent, so metadata shouJd
be expected to add value above and beyond the content it describes.

T o complicate ma tters, the distinction between data and metadata


can be flu.id . What is metadata in one context may be p ure data in
ano ther. For example, if yo u are looking for an article o n a cert'W1
to pic by a certain author, then the writer's name and the subject
keywords arc metadata and tl1e content o f the article is data. By
con trast, say you are trying to remember the name of the author who
wrote a particular article in tbe 1940s and can't remem ber the title.
You uo remember that it contained tbc pbrasc: "Man canno t ho pe
fully to duplicate th.is mental process artificially, but he certainly
ought to be able to learn from it." In th.is case the publication date
ra nge, 1940-1949, and the con tent of the article itself are the
metadata and tl1e author's name is tl1e data. 1
Cbapter Two 25

The Value ofMetadata


In late '1988, a non-descript van pulled up in front of Christie's
East, the pmchasing office of the renowned auction house in
New York City. Tied to its top with several lengths of rope was a
six by s.eveo foot canvas. T he driver had found it at a warehouse
sale of unclaimed property and purchased it on a whim for
$1,000. The painting was in bad shape and nothinK was known
about ·it, but it was large and old and ougbt to be worth
something. He offered it to Christie's for $1,500. Ian Kennedy, a
residen~ expert of Old Masters for Christie's e..~amined the
painting an instantly recognized it as a work of tbe Italian Master
D osso Dossi. With this n ew bit of information, the asking price
.rose from $1,500 to $800,000. It was purchased by the London
art deal~rs Hazlitt, Gooden & Fox for $4 million, dirt, tips and
all. Two months later it was sold to the Getty Museum for an
even higher price.

11
Allegory of Fortune, 11 Dosso Dossi
26 Metadata

The defining characteristic o f metadata is that whatever form it takes,


it facilitates the identificatio n and discovery o f a discrete package o f
information. The classic example of this is the library catalog card.
Independent of any actual content fro m the item being described, a
simple 3" x 5" card can provide a wealth o f informatio n that is usefu l
in locating and managing an informatio n resource, in this case a
book. At a glance, we can determine the title, author, publisher,
length, topic and even locatio n of the book. This quick access is by
design.

973.4
B21 UcCullough, David C .
John Ada.ms / [by] David lkCull ough
Mei.r Yo r k : Simon & Schuster, c2 0 01
751 p., (40) p. of plates : ill. (some c ol.) ,
maps ; 2 5 cm.

Includes bibliographical r efere n c es (p . 703-726)


and inde x.
ISBI-! 0 -7432-23 13 - 6

l. Adams , John, 1 735-1826 , 2. Pr esidents - United


Sta t e s - Biography. 3. Un i ted States - Po l iti cs
and govema e n t - 1783-1809 . I. Ti t le.

E . 322. H38 2001


9 73.4' 4' 092 [BJ 2001027010

Figure I. Mctadata in a traditional card catalog.

/\ n often overlooked feature of the humble card catalog is that the


cards are organized to facilitate this at-a-glance utility. E ach card has
a consistent location and format for each piece o f information it
contains. When looking at an author card, we know the first line
indicates the author of the work and the second line is the book's
title. Th e structure o f the card telJs us that a boo k is a biography of
J o hn Adams written by D avid McCullough rather than the o ther way
around. The same principle applies to electronic resources. To be
useful, mctadata must be structured to facilitate both discovery and
interpreta tio n.
Chap ter Two 27

Most major newspapers now provide onJine editions with searchable


full-text archives. Tf we type in a few well chosen key words, we have
a chance of finding something of .interest. The newspaper's search
engine will match our query terms against every word o f every article
of every edition contained in the archive. This is searching the data,
the actual content of the newspapers. This type of search is subject to
all of the pitfalls of unco nstrained search as discussed in the prior
chapter. If we instead search the meladata, we can dramatically
improve the effectiveness of our search.

111:WS f llTEllTAIIIMEllT OTHm StCTIOIIS ClASSlftEDS JOBS CARS HOMES REIITALS

• JOBS Sea,ch ror:


• CARS
-------
• HOMES Coment O11llo11s: 0 111.'1985 . Present (Te><Q
• REIITALS
MORE Cln$1FIEOS 0 121-1•1881 - 12/311198-1 (Htstonc Article Images)
SAi.ES &DEALS
8USIMEst OIRECTO~ Soll By: 0 Most Recent First
eo._,.1,..__kjf
rucE.,IID 0 Oldest Fhsl
0 Retavance
ARCHIVES
Ba~k Searc.h Date Options: 0 All dates
AdvolO~f.ld $ e.u ch
O oate Range
s.wed Search r.:: ,, ....,,~
Login F1 0 111; .wl v i~
~ ·: 1
Account & PUl'Ch.1$C$
Knowledge Ccnte, To: ~ ~ ; - - : : ~
-- L--
Arc.hive• Trouble
Rer,011
AtRhor: (optlonaQ
l.nlmea.com Sit♦
Headline: - - - ~ (option•~
Servke•
ARC:HIVES Hfl.l' ~ lllf•' A1ticle Type: Al
Abot.n the ArclW'f: Sectloo: Al- - - -
Prl<ing
Term& of Service Sem ell O1>11011s: Search Articles Only
Se,u ch TQJa
FAO Search Mieles. Advertisements and Listings
Storie,: Prio-rto 1tl3S
EIL#M V'

Figure 2. The adva nced sea rch page o f the LA Times.


28 Metadata

Tf we would like to research the position of fom1cr president Jimmy


Carter on U.S. trade with China, a reasonable place to start is the
arch.ivcs of the Los Angeles Times (www.latimes.com). J\s we would
• on th c Iceywor d s "Carter, " "]) o Licy, " an d "Cluna
expect, scarehJng • "
returns an assortment of documents ranging from an analysis o f the
conflict between China and Taiwan to an obituary of Stanford
University professor Michael Oksenberg. Fortunately, the Times
archive provides an advanced search mechanism utilizing ex tensive
meta.data. Rather than a blind search where all words are treated
cgually, the Times enables users to restrict certain terms to certain
areas. We can specify that "Jimmy Carter" only be matched against
authors and th at o nly articles of the type "opinio n piece" with the
word "C hina" i.o the head.line be retrieved. Even tho ugh we are no
longer looking at any of the archive's actual data or article text and
are instead searching only meta.data, we receive a precise set of
documents with a strong likelihood of being .relevant to o u.r interest.

Types ofMetadata
The advantages metac.lata affords to sear c hing electro nic versions o f
traditio nal textual resources are straigh tforward. However, the digital
world isn't as simple a place as it o nce was, and newspapers,
magazine articles and the like arc rapidly becoming a mino rity among
the milieu of online information. ew types of i.nformation objects
and arti facts seem to emerge daily. Io order to manage this deluge o f
new forms o f in formation, we must be able to describe them in ways
that are specific to each wuguc type and the tasks utilizing them. To
this encl, several different forms of metadata- desc.riptivc, technical,
and administrative- may be developed for any given in formatio n
object.

Descriptive Metadata
D escriptive metadata is by far the most common form o f meta.data
i.n use today and is usually what you will encounter as an in formatio n
Chapter Two 29

seeker. This type o f metadata comprises what is explicitly added to


content to make it easier to find. lo a nutshell, descriptive metadata is
the who, what, when, and where of an in formation resource. \'v'hile it
found its first broad ap plicatio n with textual resources such as the
LA Times archives, it is rapidly coming to permeate every aspect of
the online world.

Take for example, Apple Computer's popular iTunes o nline music


service. Sin ce the content offered by i'l'unes is non-textual (i. e., the
strains o f a Bach concerto or a John Coltrane solo), full-text search of
the content itself is ill-suited to retrieval. Rather, you search the
textual information associated with the audio or video file you are
trying to find. Most files have been extensively tagged with
descriptive metadata. This includes the basics, such as artist, album,
and song title as well as more advanced categories such as genre, sub-
genre, release date and publisher. Each piece of metadata associated

,:.]~
q......,_ o ,@=
......yrt, ...,...";).-« ~

Tor Attb,t Or...~ b l,i'l.t ,irtot1!1 .lll!SOl>ouglll


Th• Rloe >nd f>II ofZlggy
Stoudusr o1nd the Spider~
I . Unde, P'l 1ttsur•
2 ~ e Oddiry ,..,,,..,.,.
l.9:w.-eOddRy.

fromM3n 3 undtf P'l tt,w e AI.Mklln Saoe.


• , .,., Dance
03\id BOWie
(;if1 lhlct.tuck 0
5 Ch-M'OO•
d. Rt!bel Rebl-1
Lo,.., "'""'"""'
D,v,a-
AtlklAAttt 0
Tehflieod 0
11..-.esEt~ T~f't41.etl M:c.u
~.Deu.Sep28. 19SI(!:
t ~k.it>e- 1. ~~ :t h om~tt.ero-
~~~ SHM O 1. eu... fleiO
I l.ci.., ,kttodl
• f<02S Mro5t.F\.,. Csil H!rJ \l'fvln ol. Dl'Ad 1o.,.1,e ~•v I ow ...
lt.tt . .........,~
s:. lhrlr-.n.oa:

l , . Pt<Mo Put•
1. Ro.t , Vf rut•
•-A>-0 u
t;

iT unes Review Customer Reviews t"g '.'t1,•A■ 11Re-'Mwc O


lWR't!:)(wt,.ol 51•11\.'WI l h t r ~ t o c i ~ - ! l ' N ~ b y
enegolo01UliS ...... lo)'•&neremorblto~ lWll\l'N,llmde'S'slW
l lOl'lfl'l:ofltl6tietnYJ~.KUOl-70ttd ~ . D l ' ! M ~
~.,~ptng?",m11~~ .wtfk. ,byn ~.,,,,.., weco.Jtfaf
eu..JtA11-hBeS1 * :lf..:A *Jt
b'tJitt-....,. ,.. -
~ nrac:t.~~rncrtll'\:«ttltth::cied:a:t'i.:t
~rqCll'NdlGf~. N ~0tu:ttne-¥bf!:e,Pwt",t,:1ectl:o Trclllliftltshoa,-.e:~os•Q-'• rt oneotlhOMotuct.
,1 A.fl"'flll. ~ IOd:Rd~w~U90(3~-..wfOU ~~
.....nere 1 r 1 ~ : o p d .<:t.Aa.i:::1onc;c!OdlCW'9 boo<'ao1,etis11
bac.h "Sla,JMt,;~)e'(~.•"frf9Y•1;"1-w,QCrt:t e,M1tir1M«-.or ~•ia.a:::ill, 'Wf!Ke'!!'Cll.l<<in111dilrat..-cift.~O!t
Ytur:d!' ....«tt,.,,.ior,1,~0,:tC\iN'liel- Moire-

---
'f:42 Ob-,,d Bowie 0 lheRtse«ldfal of Z.. .
Swl= J:33 O.Vld~ 0 ft,elbwloodF.,fofz.. .
4Jl8 O.Yidao.. 0 Tt.RNondFalofZ...
◄ S(a,,no,-, 4:13 o.w,ea-, 0 TheRaundfaldZ... SOW~
5 llAn't.~JY 2:$1 Oraw!ltor,<,,'JO 0 lhoRisoMd.Falofz... 1().99 ...,
6
...
l«h'5lerclnt 1:20 O.W,Bo,,,,w!,
2:i6 [)r,,dflO'tle
Z:38 oa'Yld&ow,,e-
0
0
0
TheRISoMIOFalofZ...
TheR.Q! YiCFaldZ...
TheR.AlrldFalof Z...
S0-99 ~
$0,99 ~
S0.99C., • • o - )
.,,,,,,,._
11,q(J"ltoYo,ssef
J;l2 0r,41:UJowe 0 TI-.Ra.-.:!Falof z... MunOri)o
3:2'40.W,BtMle () TheRae.-,cFaldZ... S0.99

II~

Figure 3. Metadata in iTuncs.


30 M e tad ata

with a particular song increases the probability that it will be found,


either by searching or browsing, and subsequently sold.

The value of descriptive mcta<lata doesn't rest solely in discovery and


retrieval It also facilitates tl1e second part of the e-commerce
equation: making the sale. Once a user browses tl1rough genres, sub-
gemcs, and artists to a particular albwn of interest they can read
reviews, ratings, song length, and even beats per minute. All of this is
descriptive metadata that will help ilic information seeker make a
value judgment of the content t11cy arc considering. The principle is
equally valid for corporate earnings reports as it is for Mariah Carey
videos.

Administrative Metadata
If descriptive mctadata is intended primarily for the information
seeker, administrative metadata is 1na.inly for the benefit of tl1e
information owner or steward. Metadata elements specifying from
where a file or document came, where it is to be hosted, who is
authorized to modify it, when it is to be archived, in what form and
for how long arc all administrative mctadata. It is created for the
purposes of management, decision making and record keeping. 3

Administrative mctadata is tl1e lifeblood of modern content,


document and records management systems. It allows content to
move through its lifecycle in a largely automated fashion. For
example, companies try to keep ilieir websites interesting by
continually changing their content. cw stories arc posted to the
homepage and older content is moved to less prominent locations. J\
few well chosen pieces of metadata, such as publish date, run length,
and archive page ID can combine with business ruJcs in a content
management system to automate for tlic most part the entire process
of updating a website. This frees the Web team to focus on creating
compelling content rather tlian shuffling files around the server. I t
also allows tl, e website to be updated in the middle of the night
wiiliout disturbing the webmastcr's sleep.
Chapter Two 31

Recently, administrative mctadata has found a new niche in the form


of Digital Rights Managem ent (DRM). Once the province of
military intelligence and industrial secrets, DRM has recently moved
into the mainstream. As distribution of intellectual property across
tl1e Internet and corporate Intra.nets has become the norm, having a
reliable means to track that con tent and control who can access it has
become essential. DRM secures digital materials and limits access to
o nly those with tl1e proper autl1orizatio n. In addition, a complete
DRM solution facilitates and tracks any transactions involving tl,c
content you wish to protect. !,.or example, allowing copying o r
limiting the period of access o r the number of ti.mes content may be
viewed must all be supported. 4 ORM technologies and techniques arc
dnven by administrative meta.data.

Structural Meta data


As we have noted, informatio n comes in many forms and &o m many
sources, usually bundled into packages tl1at a.re largely black boxes to
us. How a.re we, or more importantly ilic tools we use, to know how
the information is to be read, manipulated and displayed? How docs
an application know the technical requirements for integrating the
conten ts of some strange new file into its world so that we may hav e
access to its contents? This is the role of structural m etadata.

Structural mecadata, sometimes referred to as technical metadata,


display metadata or use metadata, describes how an information
object, usually a file or set o f related files, is put togetl,er. T his can
range &o m technical details such as file size, compression scheme,
and scanning resolution to display and navigation information such
as presentation order, typographic instructions, and search
mechanism s.

The most common application of struclural metadata is defining how


in formation is to be organized in databases and data wareho uses.
Every piece of informatio n housed in a database must be grouped
into records and described in terms o f type, size, and relationships.
32 Metadata

The structural metadata governing this organization is in fact what


makes up a database and turns unorganized data into a usable
collection of structured information.

Another way of looking at structural metadata is the page-turner


model. In this model, structural metadata specifies how individual
informatio n objects are bound together to make up a single
information package that is presented in a specific order, like the
pages and chapters of a book. This allows text, images, and other
content to be presented in sequence, but enables the user to navigate
it at will, jumping from section to section, while preserving the
organization and structure originally intended by the creator.

Metadata Schemas
Regardless o f its type- descriptive, administrative or structural-and
the purpose to which it is applied, all metadata share cer tain
characteristics. At a minimum metadata must posses semantics,
synta..x, and structure .5

Semantics refers to the meaning o f metadata within a pmticular


comJtnmi!J or domain. Tt is important to note that any given metadata
field can have different interpretations depending on the context in
which it is being used. For example, the administrative field sample
so11rce could refer to a medical procedure or even a particular patient
in a medical context, or it could refer to a certain musical instrument
or recording in the context of audio productio n. It could just as easily
be a technical field referencing a particular device or encoding
scheme. The p oint is that without clearly defined semantics, it is
nearly impossible to accurately interpret mctadata.

Just as people cannot interpret metadata without an understanding of


its semantics, computers can't make sense o f it witho ut syntax and
structure. Syntax is the systematic arrangement o f metadata elemen ts
and their values according to well defined rules. The most common
Chapte r Two 33

form of syntax currently is the name-value pair in which the name of


the metadata clement is simply matched with its value, such as:

<author = Arturo Perez-Reverte>


<title = The Club Dumas>
<genre = Fiction>

Structure defines how metadata is to be organized to ensure


consistent representation and interpretation in line with its syntax and
semantics. The structure specifies which mctadata elements are
allowed where, in what order and how often. A record describing a
"book" must start with one or more authors, followed by a single
title, a single genre, an optional sub-genre, a single publisher and so
forth.

Taken together, semantics, syntax, and structure form a type of


grammar, called a schema, that specifics the rules governing the
metadata of any given domain or application. At the most basic level,
a schema specifics a list o f attributes that arc valid for describing ao
information package. A more sophisticated schema will often detail
out every aspect of how metadata is to be encoded and represented.
In all cases the overarching gmtl of defining a rich schema is to make
metadata as useful as possible in terms o f interoperability,
extensibility and flexibility.

Interoperability is the ability of information systems to exchange


metadata an<l interact in a useful way over communication networks
such as the Internet.(' This is what allows the computers at
Amazon.com to talk to your bank or credit card company and receive
payment for the book you ordered. Extens ibility means that the
original definition o f the schema isn't the final word. I t should always
be possible to add additional metadata elements (albeit in an
organized and controlled manner) to any schema in order to
accommodate specific and often L111forescen user needs.
34 Metadata

Above all, mctadata users demand flexibility from their metadata


schemes and systems. T hey d o no t want to be com pelled to add
information that they deem is irrelevant or too cumbersome. As a
result, most mctadata schemas allow authors to include as much or as
little detail as they desire in a metadata record. This makes autl10rs
happy, but tends to make life difficult for informatio n aod metadata
administrato rs, since the more flexible mctadata is, the less
interoperable i t becomes. Two informatio n systems may depend o n a
particular metadata elem ent in order to communicate, and if an
author fails to provide it, interactio n between tl1c t\vo systems
becomes impossible. Imagine if Amazon.com neglected to include
the pr ice of a book when it tried to charge your credit card. Schemas
serve to mitigate tl1ese problems while presc1v ing as much flexibility
as possible.

T he num ber of publicly available schemas has exploded in recent


years, and there now seems to be metadata standards (official, de
facto, and even competing) for nearly every domain imaginable. O ne
o f the earliest and most broadl y applied is the D ublin Core (DC).
am ed after the O hio city in which it was first drafted, the D ublin
Core was originally developed witl1 an eye to describing document-
like objects. More recently, D C metadata is beginning to be applied
to a broad range of other types of resources as well.

O ne of the strengths of D C and a prime reason for its popularity is


its simplicity. The D C schema captures the fundamental characteristic
o f an information resource in a mann er tliat is easy to create and
comprehend . T homas Baker of the German National Research
Cen ter for Information T echnology has referred to it as "metadata
pidgin for digital tour ists."7

l n its current form, D C consists o f fifteen elemen ts covcnng tl1e


basic descriptive, administrative and structural needs o f an
in formation o bject. For each clement the sch ema supplies both an
o fficial label and a concise definitio n. I ;or example creator is defined
as: "an en tity primarily responsible for making the content of the
resource." Just as with a well defined structure, clear definitio ns of
Chapte r Two 35

labels and terms arc essential to ensuring the appropriate


interpretation and application of metadata.

The D ublin Core is an example of a simple schema that can mediate


between the extremes of full indexing of raw text and highly
structured content. It provides a mechanism for capturing the
fu ndamental information necessary to describe an information
rcsow:cc without the burden of elements that may be irrelevant to a
particular community or application.

Some have perceived the spare nature of D C schema as a weakness.


While its basic nature allows it to describe many different types o f
resources, it limits the detail you can capture about that resource. For
example, the creator clement, described above, makes no distinction
between a person, an organization, or a service. This could be
essential informatio n to a particular application. Perhaps even more
troublesome is the fact that there are no constraints placed o n the
values a given element may take. For example, the subject element
can be filled with a keyword, a Library of Congress Subject Heading
or a free text descrip tion. T his lack of standard terms and values is
critical, as we shall sec shortly.

Descriptive Administrative Structural

Title Creator Date


=
Subject Publisher Type
Description Contributor Format
Source Rights Identifier
Language

Relation

Coverage
Figure 4. The current Dublin Core c lement sci.
36 Metadata

T hese shortcomings arc common to most metadata schemas. The


Dublin Core is a good example o f how linutatio ns can be overcome
through extensibility. T he DC supports two types o f qualifiers,
schemes and types, which refine the base schema.

Schemas allow you to specify the standard syn tax o r vocabulary that
arc allowable for clement values. T he D C element Slf~jec/ may be
qualified with MESH to indicate that all values must be drawn from the
Medical Subject Headings vocabulary or LCSH to require Library of
Congress terms. Likewise the language clement may be qualified with
ISO 639-2RFC 3066 to ensure that any value applied to that field
conforms to the ISO standard.

DC types refine the definition of the core element itself. The basic
D C clement date, defined as "a date associated with an event in the
life cycle o f the resource" is too generic to be useful. 13y applying a
type, the basic date clement can be transformed into date created,
issued, accepted, available, or acquired, among other possibilities.
This ability to refine and enhance the schema without corrupting its
fu ndamental nature and structure is the key to metadata extensibility.
\'qithout it, any metadata system will quickly become o bsolete
regardless of bow well conceived and executed initially.

Where Do I Put It?


Mctadata can live in several different places. Traditio oaUy, as with the
card catalog, it has been recorded and stored separately from the
object it describes with a pointer of some sort to the location of the
in formation resource itself. This is o ften the case in content
management and data warehouse systems. Information resources will
be given a unique iden tifier and stored in whatever fo rm and o n
whatever system is most appropriate. 'fhe metadata describing that
resource may be h osted in a separate database dedicated to that
purp ose. The metadata and the object it describes remain I.inked by
means of the resource's identifier.
Chap ter Two 37

This approach has the advantage of making it simple to update the


metadata o f any given information resource. I f a new manager takes
over responsibility for a large number of documents, you can simply
update the database with the new informatio n rather than tracking
down and retagging the documen ts themselves. The disadvantage o f
this approach is that the metadata doesn't travel with the document if
it is shared. If a file with externally managed metadata is ern,'liled to a
colleague at another organ ization, they will receive the content but
not the descriptive informa tio n. This can become a problem if th at
additional inf01matio n is critical to making the document usable.

Ao alternative to external management is to make the metadata a part


of the information resource itself. Most applications supporting thjs
approach store metadata as properties o f the file they describe.
Mjcrosoft Windows, for example, allows an auth or to add summary
metadata to any file, which may then be used to organize, locate, a.nd
retrieve the information resource. In additio n to traveling with the
file, internal metadata has the advan tage o f being somewhat self-
maintainiog. In the case o f Windows metadata, some information is
extracted directly and automatically from the document itself. The
organization of the file is automatically extracted from heading styles
in a Word document, Excel worksh eet titles, or slide titles in a
P owerPoint presentation. If the file changes, the new structure is
auto matically reflected in the metadata. Usage statistics are also
automatically updated throughout the life o f the document. At first
blush, semi-automatic maintenance and close coupling witl1 tl1e
in formation it describes makes internal metadata a very attractive
option, but it does come at a cost.

rirst, while some o f the descriptive metadata (title, author, cornpa,!Y) can
be au tomatically generated, the fields that are most useful to retrieval
(su~ject, category, kry111ords) must be manually selected, keyed, and
maintained. If tl1e owner o f the document changes, as mentio ned
earlier, not only docs that field need to be updated in each impacted
docwnent, tl,ere will be no history o f own ership. O nce ao internal
field is updated, all previous values are lost. This can become critical
if an explanation of something in ilie document is needed and n o o ne
remembers who origi.nalJy wrote it.
38 M e tadata
----·· . ~-- ··-- ··· ·----· - ···--· • •·--· --·· • - -· - 'I
!lntn:irtifnfon to th"; i.;;11~~11itir Wi;bt1utlini. <lo;;-1•-: {tlti l

Property
Description

[¼Title Introduction to the Semantic Web


c;:rsubject Semantic Web
[?'category Lectures
[¥ Ke\'WOrds Semantic Web, RDF, Ontology
CJ' Comments Draft of Lecture 1

Origin

[?'source
[¥ Author Darin L. Stewart
CJ' Revision Mumber 2

'-'--_O_K_....,J_]" Cancel 11-" Apply I, _Help


Figure 5. Metadata in Microsoft Windows.

t\no ther hazard is shifting terminology. The vocabulary of any


organization or community inevitably changes over ti.me. K eywords,
subject h eadings, and even category labels need to be updated to
re flect these changes. Otherwise a search engine will no t be able to
match a relevant document tagged with obsolete tenns with a guery
from a user searching with the latest buzzwords. Additionally, while
deliberate keywords arc essential to effective retrieval, as discussed in
the prior chapter, the burden of selecting, assigning and maintaining
them falls primarily on the author (wh o is invariably overworked
already). This often leads to sporadic metadata and often
idiosyncratic tags and terms. This becomes an even greater problem
in the context o f authority con trol, which we will discuss shortly.
Chap ter Two 39

Where Does It Come From?


The potential sources of metadata and the means of creating it are as
varied as the information resources they describe. Systems for
automatic gen eration exist but rarely reach an acceptable level of
quality without human assistance. Conversely, a broad application of
metadata across an enterprise o f an y si7.e is generally too tedious for
human beings working without the help of scripts, term extractors
and tagging tools. As a result, most successful metadata endeavors
draw on a range o f sources, tools and techniques depending o n the
nature of the information under consideration and the purposes for
which it is intended.

The same principle is just as applicable to creating the metadata for a


single information resource as it is to an entire collection. In most
cases, the descriptive metadat,'l will be assigned by the creator or
author of the information. This has the advantage of terms coming
from the person most familiar with the content and its original inten t.
It has the disadvantage of the metadata reflecting the biases and
idiosyncrasies of the author, whose vocabulary may no t necessarily
reflect that of her audience.

The readers may also place the information in a different context


from that originally conceived by the author. As a resuJt, it is often
advantageous to leave the creation of descriptive metadata to the
professionals. The National Information Standards Organization
(NISO) has noted that it is o ften more efficient to have indexers or
o ther information professionals create this metadata, because the
authors rarely have the time or necessary skills.R This is, o f course, an
additional line item cost, but when lifetime cost of ownership
(especially in terms of findability) is taken into account, leaving it to
the professio nals is often cheaper in the long run.

Administrative and structural metadata will often be generated by the


technical staff that prepares an information resource to be published
and distributed. The individual scanning an image or creating a digital
recording is in the best position to supply details about resolution, bit
40 Metad,1ta

rates and encoding schemes. The individual adding the resource to


the co ntent managem ent system will know when it is to be pos ted to
the website, for how long, and where it is to be archived at the cod o f
its run.

As with any budding field, there are an abundance o f tools available


to assist in the creation of mctadata. The most common (and
cheapest) is the applicatio n o f templates such as those available in
most word processing applications. In addition to providing
standardized formatting of common document types, templates can
also guide the author in p roviding basic descriptive metadata. Even if
professio nal indexers arc utilized to create the final m etadata, it is
o ften effective for the author to create a "first draft" of the mctadata
to serve as a guide. A well conceived document template can simplify
this task and improve the quality of the mctadata.

One of the challenges of high quality mctadata is ensuring that it


confonns to the appropriate sch ema. Mark-up and tagging tools can
prompt the user for the appropriate fields, requiring those that arc
mandatory for compliance to the designated schema. Once the
mctadata is complete, the tool can either embed the metadata in the
informatio n resource itself or e>-rport it to an cxtcmal mctadata
repository or database.

Extraction tools will analyze the content of an information resource


and attempt to extract appropriate terms and values for certain
metadata fields. ror structural m ctadata, this is o ften straightforward
and quite e ffective. For more conceptual clements such as subject,
category or keyword, it gets a bit trickier. Most tools rely on a mixture
of s tatistical and computational techniques to make a best guess at
appropriate descriptive metadata. In most cases these tools require a
great deal of training in terms of sample docwncnts and target
vocabularies, and still depend on human intervention and revision.
However, much like having authors take a first pass at assigning
mctadata, automated extraction tools can dramaticalJy reduce the full
mctadata burden to a more manageable one o f cleanup and
refinement.
Chapte r Two 41

"'°'"'It::!

s,..,,,,
~Je -=-~~-+• .<11!: J~ 9 . ..

ZOMSIEI.AND
0 2llMB!EIANO r,_,_..,.

-.............
,e zoMBIElAUO Tllle:1 OYORc:
~

"''""""""'
::.

on
-~R~ R
•200'3-10-2 Fl
Pl
..,,_
l;IZ°"""""d
l:.Uf ~ El
~=
Gt<~
;IZ~[BU",~I
..,,_
lll!Z~ l
:c j°"""'
~~!HDJ ~-~
L i t e ~ aZorbekindl£iJI
lalZOlrlbd.n:!!2009!
'""'"" ~~-1g~""--
~ RMdead 5""'0
t iSNlllollheOMd
~ Z~ ( 2-0ite l luUi,01 S"-'-
Q!Thf!Hill"90Ye'(R-A.Y:ed Sr,ole-Occ
~ ZorrbeL!nd l2-0c:c 'Nu1 UooSrul
~ Kd<An(Th~8\,-l""'°"°C
i jZOltlesfZartluZCll'tld ·~
............. a:int. ~ ~ - . v i • ~ - - - ,..
:;ao.... ,ie..,..i ~. . - .llllilrd Jl'OllcoJd00-=nc&,,,,...,,rto lfl.llt,o,sflt"'

~~~:~b-=J>:/t!~aj~·~ 8
net.named~ U•:1.•£~ lnc:: M:11!1 ~ r , tlll'
In. .v,dJ,OUCSI~~ l'W!IU.-hic.tnl a!Q 114t'ih
,_Nl'ft9AIIJ.,.M• S,tj d-=,,~l'IN~h.A•d'-".,_.'
~:Oll"C'Otw..11.Wltl~h'l'!lm&efllfll'IT'.aSIMoOl'ldA,hQ."II .,.

Figure 6. A metadata creation aid: Meta-X.

Metadata and Authority Control


Metadata is a hard sell. I t is expensive to create and difficult to
maintain. Executives have a tough time understanding how the
problem of having too much information to manage can be solved by
adding on yet more info rmation. Metadata is a bit of a "hair of the
dog" solution. We add a little extra information to make a lot of
in formatio n mo re usable. J\s to the expense the answer is, of course,
pay now or pay more later; sometimes a lot m ore. As discussed in the
prior chapter, a few m omen ts tagging a docum en t can save hours
bLU1ting for it l..'lter. W hen done properly, metadata initiatives n early
always generate a positive return on investm ent. Unfortunately, few
a.re done properly and most fail. A prime reason for this is a lack of
authority co ntro l.
The notion o f au thority co ntrol boils down to making sure everyon e
in volved in the creation and managem ent o f an information resource
42 M e tadata

is speaking the same language. It is the mechanism by which


consistency in o nl.i.ne systems is created and maintained. When
applied to search and even navigation, it promotes greater precision
by providing official or "authorized" forms of names, labels and
values. As part of this system, references to equivalent terms and
9
syn onyms and variants are created which dramatically improve recall.
recall. 9 For example, if the authorized term for a "non-rigid, buoyant
airsrup" is blimp there will be cross references to zeppelin and
dirigible. An information seeker searching on any of these equivalent
terms would receive information for all of them.

Th e value of authority control to metadata should be obvious. While


schemas provide structure, syntax and semantics to ow: mctadata, thry
do nothing to ensttre comistenry i11 the values assigned to the elements of the
schema. The Dublin Core may specify an element called language and
define it as, "the language of the intellectual con tent of the resource,"
but it does n othing to limit the po tential values that can be assigned
to that field. If DC metadata is being created for a□ international
news story, its language could be tagged as English, Eng. , En, American
English, British English, or any number of variants. Each is potentially
valid, but the lack of consistency turns retrieval into a crap shoot. If
an information seeker searches on English they will receive only those
information resources labeled with that exact term. Anything tagged
with ano ther term for E nglish wilJ be ignored.

The solution is to restrict potential metaclata values to an agreed


upon list of terms, so that both information creators and seekers are
speaking the same language. Io many cases, an au thoritative
vocabulary already exists and ca□ be ado pted wholesale. Io the case
of the D C language element, the International O rganization for
Standardization (TSO) Language Codes standard (ISO 639-2)
provides authoritative names and codes for languages. English would
then be consistently represented as eng, Italian as ita, Japanese as jpn
and Esperanto as epo.

If the desired granularity docs not exist in tl1e standard, it can be


expanded. D CMI actually recommends this as a best practice in the
10
case of languagcs. T he ISO standard can be used in conjunction
Chapte r Two 43

wiili ilie Internet Societies' proposal for language codes (RFC 3066),
which includes ilie more specific labels of en-US for American
E nglish, en-AU for E nglish as used in Australia, en-GB for English in
ilie United Kingdom, o r even en-GB-oed for British English using
spelling from the O xford E nglish Dictionary. T he additional
advantage of adopting auilioritative terms is the possibility of
sa.ucturing the labels to re flect relationships.

Eng(UseFor English, en, )


En -AU (UseFor Australian English)
En -GB (UseFor British English)
En -GB-oed(UseFor Bri t ish English OED spelli ng)

D espite the advantages it offers, authority control is a difficult pill to


swallow for most organizations. The prospect of giving up ownership
of terms and labels is often enough to incite turf battles in even ilie
most collegial of enviro nments. Authors feel that it is unnecessary
and even inadvisable to constrain their vocabulary in any way (though
they invariably recognize the need for such constraints among their
coUcagucs). D eciding who and what is ilie "authority" and who and
what is governed by its dictates are among the most contentious
issues in information management. If metadata is a hard sell,
authority control can turn into a sho tgun wedding.

Fortunately, it needn't be so. A balan ce can be struck between the


expressive needs o f content authors and the findability needs of
information seekers. D oing so depends o n the proper definitio n,
creation and management o f the inform.ation resources provided to
both groups. Taxono mies arc the lynch pin of this process.
Another random document with
no related content on Scribd:
Baxter and compel him to rest in England without forcing me to break my
vows."

"Your vows, your precious vows, always your vows!" she cried, in
anger and great contempt.

"Yes," he retorted instantly, "my vows, always my vows. They are


precious to me indeed, and I will beg you not to speak of them lightly."

She faced him with increasing anger. But, before she could speak,
Antonio suddenly repented himself of his sharpness.

"Isabel," he said, in quieter tones. "Think. You despise me for keeping


my vows. But suppose I had vowed my vows to you. And suppose I should
break them, for some other woman. What then?"

"I would kill her. And you too."

For a moment her wrathful excitement hindered her logical perceptions;


but as soon as she recognized his meaning she cried:

"It's different, all different! I'm real; your Bride isn't. Besides, She has
deserted you. She's run away, or She's dead. You are free."

"No, Isabel," he said. "Think again. Suppose to-day I should vow my


vow to you. Suppose your father, or someone else, should pluck you
suddenly from my side so that I could never find you again. Nay, more.
Suppose you were untrue to me and that you abandoned me. Would you
have me say: 'She has gone. I shall never see her again. To-morrow I will
seek another bride?' No, Isabel, no. If you say Yes, I shan't believe it. I
know your soul too well. Even if you broke yours, my vow would still be
there, and you would despise me for not keeping it. Am I right or wrong?"

He had unguardedly lowered his tones to a perilous tenderness, and he


was unconsciously gazing at her with the gaze she could never resist. Her
lips lost their hardness and began to tremble, and her eyelids drooped over
her eyes.
Antonio involuntarily recoiled from the danger. He knew in an instant
that his fate was quivering in the balance. His heart had bled at every harsh
word he spoke to her; and he knew that to sweep away the last shaken ruins
of his defenses, she needed only to throw herself weeping into his arms. He
knew that if she should once sob out, "Antonio, Antonio, don't send me
away," his doom would then and there be sealed.

All this Antonio knew. But Isabel did not know it. His sudden
movement of recoil stung her back into anger.

"Are you right or wrong?" she echoed bitterly. "You're right, of course.
You always are. Even when you're wrong fifty times over, you can argue
yourself into the right. I call it cowardly."

He exhaled a deep breath. The peril was past. Her scorn he could
withstand.

"I have come to the end," she cried. "The very end. Listen. You are
blighting my life, but I won't let you blight your own. Mark me well. This
place is mine. These lands are mine. I have the right to go to-night and to
set the whole abbey ablaze; and where will your work be then?"

The threat did not alarm him; but the cruelty of it, coming from such
lips as hers, cut him to the marrow. He was on the point of retorting that the
place was not hers at all, and that her father had deceived her on a wretched
point of money. But her anguish was bitter enough without this new
mortification; so he held his peace.

"I can make a bonfire of it this minute," she went on passionately. "I
hate it. How I should love to see it blaze! But I won't. And I won't sell this
place. And when I've left it on Thursday, I'll never come back till you seek
me on your knees. Never!"

Still Antonio held his peace. Isabel picked up her little bag. But she did
not turn immediately towards home. She stood awaiting his final word.
When it failed to come her indignation rose to its climax.
"No!" she cried. "I've altered my mind. I will come back. I foresee the
end. You will never seek me. You hate me. But I will come back. You'll go
on slaving, slaving, starving, starving, praying, praying, and breaking hearts
in the name of God. But I will come back. You'll succeed. You'll regain the
abbey. You'll fill it with monks. But remember. I will come back. On the
day of your triumph, I will be there. It isn't only you Southern people who
love revenge. I will be there. I will come back!"

Antonio had been silently praying for sudden grace in his own dire
need; but he ceased to pray for himself and prayed with all his soul for her.
She turned to go.

They stood facing one another as they had stood so often during these
two bitter days of their ordeal. Try as he would the monk could not conceal
his agony of holy love; and under the spell of his gaze the devil of
revengeful hate which had entered into Isabel rent her poor heart and fled
away. They looked at each other a long time. Then, in a breaking voice, she
said softly:

"Antonio. I don't hate you. I love you. This is the very last time. Do you
send Isabel away? Is it true that I must go?"

With a sharp moan of anguish and with hands thrust out for mercy he
gave his answer.

"For the love of Jesus Christ," he cried. "Go! And may the merciful God
help us both!"

He closed his eyes in desperate prayer. But God and the Virgin Mother
and the whole company of heaven seemed to have forsaken him. No light
shown, no supernal fortitude came down. Instead of a vision of ministering
angels, his mind's eyes saw only Isabel. Isabel, standing there. Isabel,
weeping. Isabel, wounded to death by his cruel sword. Isabel, hoping
against hope for his mercy. Isabel, his Isabel, rarer than gold, lovelier than
the dawn, purer than snow, waiting to dart like a bird into the nest of his
love.
He could fight no longer. Stepping one staggering step forward he held
out his arms and opened his eyes.

She had vanished.

A moment later he caught sight of her pressing up the path above him.
She was going swiftly, looking neither to the right hand nor to the left. Now
and again a ray of the sinking sun shone upon her hair, till she seemed a
queen crowned or a saint glorified.

With all his heart Antonio yearned to leap after her, to capture her like a
shy creature of the woods, and to bear her back in triumph, seated on his
shoulder as she had sat after the thunderstorm. But his limbs refused to
obey. His feet seemed to have been rooted for centuries in the granite. He
could not move an inch.

Two cypresses, which they had often halted to admire, hid her from his
sight. A groan, which he could not stifle, broke from the monk. There was
one more point in the path, one only, where she could reappear. Would she
turn round? Would she look back? As he waited, red-hot pincers seemed to
be working and worming within him as if they would have his heart out of
his body. He felt as if he were bleeding at every pore.

She reappeared. She did not turn round. She did not look back. She was
gone.
BOOK VI

"ITE, MISSA EST"

Having charged José to place himself at the disposal of Mrs. Baxter,


Antonio took the road for Villa Branca about an hour after sunrise. Utter
weariness had brought a few hours' sleep to his eyelids; but he felt unrested
and unrefreshed. By the time he reached Santa Iria fatigue compelled him to
hire a horse.

While his mount was a-saddling the monk sat musing outside the wine-
shop. What was Isabel doing? Of what was she thinking? Had she slept?
Was she truly hating him at last? Would she come once more to the
cascade?

In answer to this last question he could hardly restrain himself from


leaping on the half-ready horse and galloping off like a whirlwind to her
presence. At the moment of his leaving the farm-house, two hours before,
this all-day expedition to Villa Branca had seemed the height of prudence;
but he suddenly saw it as the depth of cowardice and brutality. She would
come to the cascade, in vain; and, later on, she would learn from José's lips
how he had turned tail and run away. Antonio cringed and burned. A
moment later, however, he knew that he had done right. She would not be at
the cascade.

"To-morrow," he said to himself, with a dull pain gnawing in his cold


and heavy heart, "I shall see her for the last time. She will make no sign.
She will say good-bye as if there has been nothing between us. Blessed
Mother of God, help us to the end!"

He took out Sir Percy's letter and perused it once more to distract his
thoughts. He read:

Dear Senhor da Rocha,—

A post just to hand apprises me of your gentility to my daughter and her


governess. The fact that I fully expected such courteous behavior on your
part does not diminish my gratitude in respect of it; and I beg you to believe
in the sincerity of my regret that I shall be unable to present my
acknowledgments in person.

I indulge the hope that a proposal which I am about to make may not be
unacceptable to you. From our mutual friend Mr. Austin Crowberry I learn
that you wished to purchase the abbey domain, but that your offers were
unacceptable to the Minister of Finances.

I have paid a deposit of £500 to the chief of the Fazenda at Villa


Branca, and am engaged to pay £300 on New Year's Day and the balance
(£2500) in five half-yearly instalments. As I have become closely associated
with an enterprise which will involve my residing alternately in Lisbon and
London, I should find it convenient to transfer to yourself my whole bargain
as regards the abbey. That is to say, I forfeit the £500 already paid and
leave you to find £2800 on the dates above referred to. I also ask your
acceptance of the larger articles of English furniture recently placed by me
in the guest-house, and I have instructed Jackson, my man, to bring away
personal luggage only.

As my movements are erratic, perhaps you will indulge me by


completing the business with my agents, Messrs. Lemos Monteiro and
Smithson, Rua do Carmo, Lisbon, who have written to Villa Branca
preparing the officials for your visit. Failing your approval I will make
other arrangements; but, meanwhile, I beg that you will add to your
unfailing kindness by taking care of the keys, and that you will believe me to
be

Your obliged and obedient servant,


Percival Kaye-Templeman.

Once in the saddle, with the well-beloved music of horse-hoofs in his


ears, Antonio found it easier to abstract his mind from bitter thoughts. He
applied his whole brain to problems of finance. Two thousand five hundred
pounds in two years and a half. At first it had staggered him; but he was
going to take the risk. His own and José's hard cash hoardings would pay
the New Year's Day instalment nearly twice over. By mortgaging the farm
and the sea-sand vineyards, and by pledging his personal credit he could
pay the July five hundred and keep two or three hundred towards the
instalment due the following January, making up the balance from the year's
wine-sales. Fifteen hundred pounds would remain payable; and this sum he
hoped to raise in due course by a bold stroke involving a mortgage on the
abbey itself.

The chief of the Fazenda received his visitor effusively. This time the
monk was not required to lean against a pile of stolen books. He sat in the
chief's own chair and was offered wine of the chief's own stealing. As three
hundred pounds of Isabel's money had stuck to the chief's fingers the great
man was more than willing to accept Antonio in Sir Percy's place; for he
had just learned that the Englishman would be unable to meet his
obligations, and he was mortally afraid of a reopening of the transaction in
Lisbon. He even threw out mysterious hints as to further concessions which
might be arranged. Antonio listened attentively. His conscience allowed him
to plan the outwitting of the Portuguese Government as regards money
which was not honestly theirs. But as soon as he perceived that the official
was bent on more pickings for himself the monk became obtuse. He was
not willing to assist any man in the work of more completely damning his
soul; and, although Antonio clearly foresaw that he was making an enemy
and preparing sore troubles for himself in the future, he steadfastly held out
against temptation.
The autumn day was drawing to its twilight when Antonio, having given
up his horse at Santa Iria, trudged up the path to his own door. Half the way
home Isabel had queened his whole mind. On leaving Villa Branca he had
sought to preoccupy himself with the most complicated arithmetic; but,
little by little, Isabel had reclaimed her empire. As he mounted the doorstep
his heart thumped heavily. Had she written? Had she sent a message by
José? Or, most terrible and beautiful possibility of all, would he find her
sitting in the house, as in her rightful place?

He entered. There was no Isabel enlightening the dim and cheerless


room. He hurried to the table whereon, José was accustomed to leave the
letters. There was nothing. His heart chilled and shrank. Still, there was to-
morrow. Yes. He was certain to see her to-morrow.

José stamped in noisily and handed Antonio two keys.

"They have gone," he said.

So sharp a blade of anguish pierced his soul that Antonio let the keys
fall on the brick floor.

"Gone?" he echoed. "Who? When? Why? Where?"

"The English senhoras," answered José. "They started about three


o'clock, to Lisbon."

Antonio sank down upon a coffer. He had used up the last of his
strength in tramping from Santa Iria, and he had eaten nothing all day.

"I don't understand it very well," continued José. "I reached the guest-
house at half-past eight. I thought they weren't to leave until to-morrow. I
worked under the Senhor Jaxo. He didn't hurry himself at all. Joanninha
brought us cold meat and white bread and strong wine. Joanninha is the
cook. She has the longest tongue, your Worship, in Portugal. She made me
angry, talking about your Worship."

"About me? How?" asked Antonio. He felt sick and faint.


"She heard me say that your Worship would attend the senhoras to-
morrow morning. She said: 'Where is his Excellency to-day? I suppose he's
gone to see Senhor Jorge's Margarida.' I said: 'No, his Worship has
something better to do. He has gone to Villa Branca to mind his own
business, and it would be a good thing if everybody else would do the
same.' There was an English servant in the room, called Ficha. She's maid
to the Senhorita Isabel. Joanninha translated to her what I'd said, and they
both laughed, and I was very angry."

"What has this to do with the senhoras going away in such a hurry?"
asked Antonio. But, even as he finished putting the question, his own fears
supplied the answer.

"It's nothing to do with the senhoras hurrying away at all," said José
humbly. "I beg your Worship's pardon for repeating such nonsense. All I
know is that some bells rang and the Senhor Jaxo went out, and when he
came back he was in a great rage. Joanninha told me that the Senhorita
Isabel had decided to go to her illustrious father at once, and that nobody
dared oppose her."

"Did you see the senhoras? Were they well?"

"I think they were well, because I heard them quarreling," José
answered. "The dark senhora, the old one, has a temper that made me
tremble, your Worship. They went away, the senhoras and the servants in
two old shut-up carriages, but they are going to hire a better carriage on the
way. I saw the old senhora, when she handed me the keys. She sent you a
long message, but I don't think Joanninha could translate it properly. So I
asked would she write, but she didn't. They locked all up and gave me the
keys. Then they went away. They didn't say when they will come back. I
think, your Worship, that they are all mad."

"José," said his master, after a long silence, "I have eaten nothing all
day. Let me break my fast. Afterwards I have something to tell you. Prepare
me what you can while I change my clothes."

He climbed the steep and narrow stairs painfully. His cold tub revived
him, and his old clothes gave him ease. But, as he lifted his worn cloak
from its hook, the wound in his heart burst open afresh. He remembered
how often Isabel had sat, in all her daintiness, upon that same cloak's clean
but rusty folds; and how, on her own confession, she had "cried and cried
and cried like a baby" at the sight of its threadbareness.

By the time he descended José had grilled two small trout and was
placing a bottle of good white wine upon the table. Antonio's heart was
wrung anew at the thought of the simple fellow's unfailing devotion. Isabel
had come and had gone; but José remained, loving and serving his strange
master with a dumb love passing the love of women. The monk forced his
faithful disciple to sit down at table with him and to take his fair share of
the dainty fish and the animating wine. When they had finished eating and
drinking he said:

"José, I have been a good deal in and about the guest-house and the
abbey since we saved the azulejos, and many strange things have happened.
The end of it all is this. Here are the keys of the guest-house. Upstairs, in
the green box, I have all the keys of the abbey. To-day, as you know, I have
been to Villa Branca. We are in legal possession of the abbey domain, and
everything in it. Within three years we must raise three thousand pounds.
With God's help it can be done. The English people will never come back."

He closed his eyes wearily. When he half-opened them he saw José by


the light of the one candle, bowing his head and silently repeating thankful
prayers. The monk quailed. For himself, as well as for José, this ought to be
a night of praise and rejoicing. Yet Antonio found it the darkest hour of his
life. The abbey keys seemed no more than a few bits of metal. Or, if they
were more than bits of metal, they were the keys of a prison, the keys which
were locking Isabel outside his life.

He took his candle and went to bed. But, despite his weariness, he could
not sleep. Where was she? In what rough inn, amidst what discomforts and
indignities, was she lying? If he jumped up at once and tramped southward
until he could find a horse, when would he overtake her? To-morrow, he
calculated, about noon. He imagined himself thundering after her chariot,
like a highwayman in a picture. He pictured her pretty alarm, her radiant
joy, her gracious forgiveness, their ecstasy of reunion.
Suddenly the monk remembered with a shock that he had not said all his
Office. Busy or idle, sick or well, glad or sad, he had never failed to recite it
before. He still had None, Vespers, and Compline to say. Lighting the
candle and opening his breviary he began to repeat the holy words. But he
had not uttered half a dozen sentences before he shut the book with a snap.

Half an hour later he arose, put together all the keys, and went down
stairs. The new moon had not set, and its brightness lured him forth from
his narrow room into the peace of the night. As a matter of course he took
the path to the abbey.

Although the ruts of wheels, her wheels, made him shiver he did not
turn back. He opened the chapel with the long key she had so often handled,
and sitting down in his old stall, he tried to say the rest of None; but a white
form, her form, hindered him, and a soft, glad voice, her voice, cried:
"Antonio, Antonio, Antonio—what a beautiful name!" He groped his way to
his own cell, and he could almost see and hear her opening his cupboards.
He hastened through the cloisters and escaped into the wood by the secret
door.

Some dead leaves fled before him, their tripping sound was no lighter
than the fall of her elfin feet. The moon suddenly peeped at him through a
clearing; and he saw her moon-white shoulders. The chirrup of a brimming
brook struck upon his ear; and he seemed to be carrying her once more in
his arms, while she murmured: "Listen, Antonio, all the world is singing."

He knew that the guest-house must tear his wound wide open, and that
he ought to hurry home to the farm; but an irresistible influence drew him
on. He reached the broad path. He stood under the casement whence she
had flung the white rose. It was still ajar.

He turned the key in the lock and entered the ghostly and silent house.
There was enough moonlight in the salon to show him the blue ottoman
whereon she had so often sat. He hurried out of the room with a heart ready
to burst.

At the foot of the stairs he paused. They led to her chamber. Could he
bear to cross its threshold, to lean out of the window as she had leaned out
after the thunder, and to look at the bed where she had lain sobbing for his
sake? He knew he could not bear it. But his intellect had ceased to govern
him and he ascended the stairs.

A broad moonbeam lit up every corner of her chamber. Like a man


dazed he lurched to the window. There were the roses and there were the
thorns. He turned to gaze at her couch. The fine linen had been taken away;
but there was the place where she had lain, there was the pillow which her
golden head had pressed. What had her last night been? Had she hated him
or did she love him still? Had she cursed God or had she prayed?

For a moment his mind turned the question over in a numb, impersonal
way. Then he came back with a rush to himself and, in a single moment, his
chalice of agony welled up and brimmed over. He flung himself down on
his knees and stretched out desperate hands and hungry arms across the
narrow bed.

Although long minutes passed his dry-eyed, stony anguish remained.


But at last his inward, spiritual man spoke. Was he committing a grievous
sin? Was he breaking, in spirit, a vow which he was only keeping in the
letter? Had he forsaken the Creator for a creature?

Slowly, but very surely, his conscience framed the answer. No, he had
not sinned. In all his desire of her there was still nothing of the carnal mind.
He was racked and scorched by anguish, not because he had lost her love,
but because he had been forced to break her heart by refusing her his own.
She was a child, a poor lonely child with neither man nor woman to love
her, nor any God to console her; and he, Antonio, had flung her back into a
still blacker frost and sharper famine, to pine and wither without love and
without faith. Yet, in all this, he had simply obeyed God. He had obeyed the
God who commanded Abram to offer up Isaac, the God who "spared not
His own Son, but freely gave Him up for us all."

The moonbeam softly faded from the chamber. But Antonio did not
move. His weary limbs and exhausted brain could resist no longer; and, still
kneeling against her pillow with his arms outstretched across her bed, he
fell asleep.
II

When the monk awoke day was dawning. For a while memory failed
him. But as soon as he understood that he was in Isabel's room he leaped up
and hastened downstairs.

He knew that he ought to go straight home. But his feet, despite their
soreness, turned towards the stepping-stones. He retraced the path by which
she had left him, hardly thirty-six hours before. Past the cypresses, through
the mimosas, he went; and before the sun rose he was standing in the icy
spray of the thunderous waterfall. He longed to plunge into the crystal pool;
but her invisible presence abashed him, and with an ever-sharpening pain he
hurried away.

As he regained the farm, he found José burning some dead leaves. Why
could he not tear down these clinging memories of Isabel from his heart, as
José could tear down ivies from the trees, and fling them a-top of the
glowing, fuming pyre? The gust of pale, acrid smoke which nipped his
nostrils was bitter-sweet.

After a dip in the brook he drank some of the sham coffee and forced
down a hunk of coarse bread. But when he faced his routine he found that
he could neither work nor pray. The black and red letters in his breviary
danced impishly before his eyes; and when he took up a pen to write out
some accounts he marked the paper with more blots than figures. Both door
and window were wide open to the morning breeze; yet the room suffocated
him.

At last a plan formed in Antonio's brain and he did not delay its
execution. Stuffing a piece of bread in his pocket he sought out José and
said:
"To-morrow my hard work will begin. To-day I am going to Navares.
After to-night I will not leave you so much alone."

He set out, striding northward with long strides. Every stride was a
symbol of his renunciation; for he knew that by this time Isabel would have
left her inn and that every moment was taking her farther southward to
Lisbon. On he pressed. As landmark after landmark came in sight a flood of
old memories diluted his bitter potion of new-brewed sorrow. He lived over
again the afternoon of his dusty march from the monastery amid a throng of
monks and soldiers and the evening of his solitary return. But not for long.
An hour before the white houses of Navares shone in the morning sun
Isabel had once more become the sole tenant of his mind.

The doors of the Navares' corn-factor's granary, where the monks had
held their council, were wide open; but Antonio did not pause to look
inside. As on the night of his flight, he hurried through the town and only
rested when he came to the knoll where he had bivouacked twice before.
Thence, after munching a little bread, he took the short cut through the
maize-fields to the village of the old cura; for the old cura's grave was the
goal of his hasty pilgrimage.

By an irony of fate a rustic wedding had drawn the whole population to


the church and churchyard. Their mirth so mocked the pilgrim's mood that
he had a mind to go away. But he mixed with the throngs until his
resentment at their gaiety was turned to thankfulness for the excess of
human joy over human sorrow. At last a horn was blown from the door of a
neighboring barn, and the crowd swept out of the churchyard like
stampeding buffaloes.

The plain grave of the old cura lay in a sheltered corner on the north
side of the chancel. Pious hands had brightened it with a yellow and purple
nosegay that very morning. Antonio did not kneel down. He simply
uncovered his head and strove to pray. For five minutes it was like chewing
chaff. Some devil whispered in the monk's ear that his errand was not only
silly, but in doubtful taste. The old cura was a saint, no doubt; but what had
so rough a diamond to do with so soft and lustrous and exquisite a pearl as
Isabel? Thus spake the devil, but Antonio refused his ear. Knowing that
prayer comes with praying, he prayed on.
Not until he had replaced his hat on his head and was about to go were
his prayers answered. But when the answer came, it was an answer indeed.
It almost struck him down, like the great light which struck down Saul on
the way to Damascus, and he was forced to lean against the church wall. It
was an answer which both healed the worst of his grief and showed him the
most of his duty in a single flash. It thrust into his hand a golden key to the
whole mystery of Isabel, past and future.

Like a man whose shoulders have suddenly been eased of a burden he


swung out homewards, holding his head high. Without knowing it, he
talked to himself aloud, uttering broken phrases of hope and thankfulness.
Yes, he had found the key, the master-key to all that had happened. As he
strode along he recalled his association with Isabel from the beginning, and
there was no lock his key did not fit.

Even the problem which had tried his faith most sorely was solved. In
confiding to him her story of the mysterious influence which he had begun
to exercise over her, four years before she saw his face, Isabel had declared
that their lives were interfused in an irresistible destiny. She had spoken of
this as a fact more undeniable than the sun and moon. She evidently
believed with her whole soul that God's hand had brought them together.
Yet Antonio, all through her pleading, had remained more persuaded than
ever that the selfsame God had called him to the celibate life. And the
apparent impossibility of reconciling these two equally clear, equally honest
convictions had kindled a fiery ordeal for the monk's faith. The only way
out of it seemed to be that all inward illumination was a delusion—totum
corpus tenebrosum, "the whole body full of darkness"—and that perhaps
there was no Divine Enlightener at all. But this wonderful new thought
which had come to him at the old cura's grave explained everything. He
thrust it into the most complicated wards of his spiritual doubts, and it
turned as smoothly as the damascened key was wont to turn in the lock of
the chapel. The doors of Isabel's soul rolled open before his eyes, and a
bright light shone into the furthest cranny.

As for his duty to her in the present and in the future, he understood it
no less certainly than he understood her chaste love for him in the past.
And, as soon as this duty was plain, he made haste to begin doing it; for it
was a duty of prayer, of specific, faithful, heroic, loving, unceasing prayer.
He prayed as he walked, with increasing exultation.

So rapt was he by his holy work that Antonio hardly noticed the
difference between the dusty, lonely road and the cobbled streets of noisy
Navares. He pressed southward without a pause. Was he not going home?
After a day and a night of banishment had not the farm once more become
the tranquil home of his body, and had not the chapel once more become the
rapturous home of his soul? He strode the last long league of his homeward
journey as if it had been the first; and when he met José at the gate his face
was shining like an angel's.

True to his word, Antonio rose early the next morning and threw
himself body and soul into hard work. Now that the abbey domain had
come under his care, there were hundreds of things to be done. As the sunny
and well-drained slopes were exceptionally suitable for the culture of a
profitable amber-colored wine, Antonio decided to double the area of the
monk's old vineyard immediately. In order to effect this extension and to
repair the damage done by seven years' neglect, it became necessary to
engage nearly a score of helpers, half a dozen of whom would have to be
retained in permanent employ. José, with one resident laborer, continued to
live at the farm, while the monk quietly resumed occupation of his own cell
in the monastery.

On Mondays, Wednesdays, and Fridays, Antonio dined at the farm with


José; and on Tuesdays, Thursdays, and Saturdays, José dined with his
master beside the stream in the monastery kitchen. At these week-night
meals, the conversation was usually a review of the day's operations and a
debate as to the work of the morrow; but on Sundays, when dinner was
eaten ceremoniously in the guest-house, such topics were not mentioned,
and the talk was of the great world's doings as chronicled in Antonio's
English paper, of Portugal's troubles, and, above all, of churchly and holy
things.

Not only during these Sunday talks, but also throughout their work-a-
day intercourse, José was conscious of a change in Antonio. Hitherto, the
monk had simply accepted the shaggy fellow's dumb affection; but, after
the day of his visit to the old cura's grave, he began to show that he requited
it as well. The last remains of his aloofness vanished, his speech grew
gentler, and he became more watchful of José's health and comfort. Nor was
the monk's manner changed towards José alone. In all things and to all
persons he was more tender and less cold.

On the long winter evenings the two men busied themselves with blue
pigments and white glazes, until they succeeded in fabricating tolerable
copies of the two broken azulejos. When this was achieved, they began a
series of experiments, with a view to distilling a new liqueur from
eucalyptus. By rashly gulping down a mouthful of the first pint, José almost
burned out his tongue. Nevertheless, they persevered; and, in the long run,
the monkish talent for cordial-making enabled Antonio to mollify the
harshness of the fiery elixir, and to render it palatable. In January they
shipped samples to agents in fever-cursed regions of Spanish America, and
offered to supply the liqueur in bulk at a high price.

Meanwhile, Antonio was waxing stronger in faith, and hope, and love.
Every day he recited the whole of his Office in his old stall, sometimes with
José's assistance, sometimes alone. He began also to hear Mass in the
village church every Wednesday and Friday, and to say the whole rosary
every Sunday afternoon. In meditating on the fifteen Mysteries, he
habitually applied them to the case of Isabel; and, somehow, these thinkings
never became trite or stale. In pursuance of his plan for Isabel's well-being,
he redoubled his prayers, and offered half his Mass-hearings and
communions with the same intention.

The winter passed and the spring came; and still he had not heard a
word from her or about her. Sometimes a memory of her would suddenly
overwhelm him. When he dined at the farm with José there seemed to be
always three persons, not two, at the table. He felt that she was sitting at his
right hand, where she had sat when he gave her the painted bowl; and so
strong was his sense of her presence that he would often halt in the midst of
a sentence, as if to ask her pardon for the dryness of the talk. After the
morrow of her flight, he never visited the stepping-stones, although he
repeatedly gave José minute instructions for the conserving of the pool's
beauties. As for Isabel's chamber, he locked it up, and never re-entered it.
Yet, in spite of this reverence for everything she had touched, he never
moped or repined. He confided Isabel, as he had confided the fate of the
abbey, to the might and love of God.

When July came, he made a novena in honor of Saint Isabel, the holy
queen of Portugal, whose silver shrine was the glory of the Poor Clare's
great convent opposite Coimbra, on the heights above the Mondego. And in
August he received a long letter from young Crowberry. Seven of its eight
pages were concerned with England's theological and ecclesiastical affairs:
but in the midst of the page devoted to personal matters, the young man had
written:

Of course, you know that Isabel has taken her father to live at
Weymouth. I never see them; but I hear they are both well, and that Sir
Percy has become quite reasonable and docile. Have they told you how she
put her foot down and sent away that Excellent Creature, Mrs. Baxter? If
she hadn't pulled up Sir Percy I'm told he would have died. Now, what did
you really and truly think of Isabel? Did you see much of her, or did she
sulk? Tell me when you write.

Antonio wrote a long letter in reply; but he did not tell young
Crowberry what he really and truly thought about Isabel, nor did he so
much as mention her name. His novena was answered. It was enough for
him to know that Sir Percy lived, and that she was well.

The grape-harvest in September was a good one, and it was only by


cutting an hour from his sleep-time that the monk could fill full his
appointed measures of work and prayer. Then came October, with its
vintage of memories. On the anniversary of Senhor Jorge's serao Antonio
could be serene; for Margarida had just been happily married to a handsome
and honorable young man of Leiria, the son of a prosperous builder. But
with the approach of the anniversary of his first meeting with Isabel he
grew troubled; and, to divert his thoughts, he departed hurriedly for Lisbon,
where he had business to transact with the shippers of his wines and
cordials. In Lisbon he learned that a journey to England would be to his
advantage. But England meant Isabel; so, on the anniversary of her flight
from the guest-house, he turned his back on the capital and hastened home.

By mortgaging his farm the monk succeeded in paying the third


instalment of the abbey's price. He faced the New Year with less than
twenty pounds of ready money, and with the obligation to find five hundred
by the first of July. A request for a more flexible arrangement was flung
back at him by the Fazenda official with vindictive contempt. As the spring
advanced, Antonio laid his plan for the immediate outright purchase of the
abbey on a fifteen hundred pound mortgage before four separate persons;
but without exception they either could not or would not entertain it. In
these circumstances he felt bound to cut down his gifts to village charities
and his bounties to the hangers-on of the countryside. As a result, José came
home one day with a black eye, received while he was punishing three
village loafers for calling the Senhor da Rocha a skin-flint and a miser.

By May-day Antonio's sales of stock and the pledging of his credit had
brought him in only three hundred pounds, and there was nothing left that
he could pawn without crippling himself hopelessly in the near future. But
he was not cast down. He was doing his utmost, and he calmly left the rest
with God.
III

Very early one morning, at the end of May, Antonio heard light
footsteps passing his cell. Although he sprang up immediately from bed he
could not open his door in time to see the intruder's face or form. He caught
no more than half a moment's glimpse of a slender and darkly garbed figure
disappearing round the angle of the corridor.

Having scrambled into his clothes, he started in pursuit. The light tap-
tap of shod feet on the stones told him that his visitor was making for the
chapel. The monk, who was barefooted, followed noiselessly.

Peeping into the chapel through the little door amid the azulejos,
Antonio saw a tall spare man kneeling before the altar. Even if his back had
not been turned to Antonio it would have been impossible to see his face,
because he was hiding it in his hands. The stranger wore a long black cloak,
uncomfortably thick and heavy for the torrid Portuguese summer. But it was
plain that he did not find it too warm. With long, thin, death-pale hands, he
drew its folds more closely round his body; and, as he did so, the familiar
movement revealed his identity to Antonio.

It was Father Sebastian.

Antonio hurried forward and knelt at his side. But Sebastian did not
move, nor did he cease praying for four or five minutes; and when at last he
turned towards Antonio it was without the slightest sign of surprise. Rising
painfully, he left the altar and made a gesture, inviting Antonio to follow
him.

As Sebastian had stood next to Antonio in juniority among the choir-


monks, the stalls of the two men were side by side. Sebastian sat down in
his old place and Antonio did likewise. The chapel was dim; but the
younger man could see that the elder's body had wasted almost to a
skeleton. Yet there was nothing repellent about him. The bloom on his
cheeks and the fire in his eyes had the solemn beauty of a sunset in an
autumnal forest. When he began to speak his voice was so soft and sweet
that it seemed to come from some far-off holy height.

"To-day, Father Antonio," he said, "completes the ninth year since you
sat on the cloister roof and heard the hoofs of the horsemen who had come
to thrust us from this house. And, this morning, it is just nine years since
you were raised to the priesthood. I asked our Lord to give me strength for
the journey, so that I might spend this anniversary with you. He has heard
me."

"Who told you that I was here?" Antonio asked.

Sebastian did not reply. But there was that in his eyes which gave
Antonio a sufficient answer. Here was a saint who walked in the light.

"Nine years," mused Sebastian aloud. "And you have not yet said your
first Mass."

"No," replied Antonio. "But God is good. Every year He enables me to


send a little cask of wine for the altar to a poor church in England. Six days
a week I work amid wine; and is not wine the matter of His great
Sacrament? It consoles me to know that although I cannot say Mass, I can
serve His table. Although I cannot, like Mary, his mother, bear Him in my
hands, I can be like those other Marys at the sepulcher. Emerunt aromata ut
venientes ungerent Jesum: 'They brought sweet spices that they might
anoint Jesus.'"

"He is not a God of the dead, but of the living," said Sebastian, in sweet,
far-off tones. "We do not offer a dead Christ. Say rather that you are like
that favored unknown to whom He sent two disciples saying, Ubi est
diversorium ubi pascha cum discipulis meis manducem: 'Where is the
guest-chamber where I may eat the Passover with My disciples?' But come.
Our time together is short, and there is much to say. First of all, I have
brought your breviary which you charged me to keep."

He pointed to a package lying on the Prior's seat. Antonio rose and took
it with joyful gratitude. When he returned to his stall he said:
"Suffer my questions first. Whence do you come? Where have you lived
these nine long years?"

"For a few months I was with the English fathers in Lisbon," Sebastian
answered. "They were kind; but when it became plain that the Portuguese
Benedictine congregation must come to an end, I crossed Spain and sought
asylum at the Montserrat, where men used to believe the Holy Grail was
treasured. There was much work for me to do there in the School of Music;
and I found strength to do it, for we lived like eagles high up in the pure air,
three thousand feet above the sea. But Madrid followed the example of
Lisbon. Greedy eyes were cast on our possessions. They accused us of
being Carlists, just as in Portugal they accused us of being Miguelistas: and
only eighteen months after leaving this abbey, I was again an exile. Since
then I have dwelt in three religious houses; and every one of them has been
suppressed."

"Can it be," asked Antonio uneasily, "that the Orders are themselves to
blame, as men say? Here we dwelt in simplicity and piety, living by our
own labor and feeding the poor. But was this house an exception? Had the
majority of other monks indeed sunk into gluttony and sloth?"

"In every monastery from which I have been driven," said Sebastian,
"our evictors poured regrets and compliments upon us. It was always the
misdeeds of 'others,' for which we had to suffer. But whenever I questioned
an exiled community, I found they had received the same compliments.
Those mysterious 'others' have still to be found. According to the statesmen,
all religious houses individually are fountains of light and blessing to their
neighbors; but collectively they are a dark curse on the nations."

"Unbelieving men are determined to mulct us of all we have," said


Antonio, "and therefore they must needs invent crimes to suit our
punishment. They hang us first and indict us afterwards."

"They oppress us," agreed Sebastian, "in the great and sacred name of
liberty. But the avarice of godless men is the mainspring of it all. I have
seen five houses confiscated 'for the good of the People'; and in not one
case have the People received a third of the plunder. But enough of this. Tell
me your own story."
"Where is the Prior?"

"He is dead. He died in Belgium."

"The Cellarer?"

"He is dead. He died in Brazil."

"Father Isidoro?"

"He is dead. He died in Spain."

With a sinking heart, Antonio named the choir-monks one by one; and,
after each name, Sebastian answered: "He is dead." Father Sebastian
believed that Brother Cypriano was still alive; but, of the Fathers, only he
and Antonio were left.

"Requiem aeternam dona eis, Domine," murmured Antonio.

"Et lux perpetua luceat eis," Sebastian responded.

"To-morrow morning," added Sebastian, "will be the anniversary of the


Abbot's last Mass. If our Lord will give me strength I shall say Mass at this
altar once more."

After a pause, Antonio began to relate his history from the moment of
his quitting the council at Navares. Every fact that threw light on his
operations for regaining the abbey he stated with precision. But he did not
mention Margarida, and he referred to Isabel only as Sir Percy's moneyed
daughter. When he had finished, Sebastian looked at him with steadfast
pitiful eyes and said:

"These have been great sacrifices and cruel hardships for the sake of our
Lord, and they will not be in vain. But you have not told me all. My brother,
I feel that you have kept silence concerning your most costly sacrifice, your
bitterest ordeal. Why not tell me all?"

Antonio's pride rebelled. The desire to ease his heart by pouring out its
hoard of solitary grief was strong; but his gentleman's instincts of reticence
were stronger. For some time he remained silent. But an inward voice
sternly bade him speak; and he spoke.

He told the short tale of Margarida. Then he unfolded the whole case of
Isabel, glossing over nothing. He scrupulously added an account of his
actions and feelings on the night and morrow of her flight. When he had
finished he sat with bowed head and waited for Sebastian's judgment. But
Sebastian remained silent.

"You do not speak," said Antonio. "Perhaps I have given you the
impression that my ordeal was carnal, and that this English maiden was a
direct emissary of Satan. If you think so, I have spoken blunderingly
indeed."

"Satan exists and he is busy enough," returned Sebastian. "But in trying


to find the cause of any strange thing that happens I have learned to think of
Satan last. Nearly all our temptations arise from our own self-love and
carelessness. Many other temptations are God's provings and perfectings of
our spiritual mettle. Satan is not omnipresent and his angels are only a
shrunken legion. But have patience. Let me think."

He resumed his meditation. At last he turned and said:

"This was not a temptation from the devil. Neither did it spring from
corruptness in your heart or in hers. I am persuaded that our Lord's work is
somehow in it all. Perhaps you will never know in this world what work it
is; but that is not your affair."

"Sometimes," said Antonio slowly, "it troubles my conscience. As I told


you just now, I didn't hold out to the very end. I gave way within my heart;
but when I opened my eyes she had vanished."

"You do wrong to be troubled," said Sebastian. "You held out to the


bitter end of the trial God had appointed you. When you told this Isabel
finally to go, she went. That was the end. All that happened afterwards was
mere reaction."
"The next day," persisted Antonio, "I did not say my Office. My heart
bled for her as it never bled for the Abbot, or for you, Sebastian, or for this
place."

"It bled for her, not for yourself," Sebastian explained. "In profane love,
the lover who thinks he is grieving for the beloved is only grieving over his
own loss of her, over his own short bereavement, or over his own
humiliation and discomfiture. With you, Antonio, it was not so. You did not
wish to take; you wished to give."

"Do not make me out a saint when I know I am a sinner," said Antonio,
almost sharply. "If she had been old, and tart, and ugly, would my heart
have bled for her all the same?"

"Perhaps not," Sebastian retorted. "But, if she had been old and ugly,
neither would there have been much virtue in giving her up. Do not
complain of her beauty. You had heroic work to do, and her beauty helped
you to do it better. In England there are Puritans who would say that these
azulejos and these gilded carvings must hinder us from doing the Work of
God."

"I do not follow you," said Antonio.

"Tell me," Sebastian asked abruptly, "how you stand with the payments
you have bound yourself to make."

Antonio drew from his breast an account over which he had pored and
pored for a month without making the adverse balance a vintem less.
Sebastian conned it attentively from beginning to end. Then he said:

"Follow me to my old cell and bring me paper and ink."

He rose with so much difficulty that Antonio had to support him; but
once fairly on his feet he moved quickly over the pavement. At the door of
the cell Antonio left him; but before he had finished cutting a new quill and
replenishing the sand-sprinkler in his own room, Sebastian rejoined him.
Sitting down painfully at the tiny table he swiftly wrote a very short letter.
Without reading it over he folded it, sealed it with a small brass seal which
he drew from his pocket, and addressed it to a Spanish nobleman in a small
town of the Asturias.

"Let this be despatched at once," he said. "There is no time to lose."

"A post leaves Navares in three days," replied Antonio. "José shall take
the letter there this morning."

"It is well," said Sebastian. "And when this José returns, let me see him
as soon as he is rested."

The cell was brighter than the chapel, and Antonio perceived that his
friend was become almost as insubstantial as a ghost. He called to mind a
passage from a new English poet about a man who, having wasted to a
shadow, was ready to be resumed into the Great Shadow, the shadow and
blackness of death. But Sebastian seemed rather to be a pure white flame,
waiting to be drawn into the Great Light.

"You have not broken your fast," cried Antonio in shame and alarm.
"You must eat. I have good wine. You must rest. You must sleep. When the
heat is over we will talk again, and you shall see José."

"It has been meat and drink and rest and sleep to see you again, Father
Antonio, and to hear what you have told me," the other answered. "But you
are right. I must sleep. I will obey your orders."

At breakfast Sebastian ate and drank nothing save an ounce or two of


bread and an egg beaten up in white wine. When the meal was over he
declined Antonio's pressing offer of a comfortable bed from the guest-
house, and lay down on the straw mattress in his own cell. There he soon
fell into so profound a sleep that he did not hear Antonio drenching the
window with bucketful after bucketful of water to counteract the blazing
heat.

At night José, wearing his best coat and his most diffident manner,
dined with the two monks in a corner of the refectory. Sebastian, with bright
eyes and glowing cheeks, did most of the talking. He praised the wine and
the food, although he touched little of either; and throughout the repast he
was full of an eager cheerfulness such as Antonio had never seen in him
before. After dinner he drew from José an exact account of his mental and
spiritual state: for Antonio had told him of the poor fellow's desire to
become a monk.

"José," he demanded, at the end of his questioning, "You have learnt


Latin. Can you translate, Irascimini: et nolite peccare?"

"I can, Father," answered José proudly. "It means, 'Be angry and sin
not.'"

"So it does. You did well to be angry with the greedy and lazy good-for-
nothings who spake evil of Father Antonio. But you did ill to thrash them
and to come home with that black eye. Go on being angry with sin; but
learn to love sinners."

"Can't I be a monk, Father? May I not have the habit?" pleaded José, in
consternation. "I am glad I thrashed them; but I'm sure I shan't need to
thrash them again."

"The habit is a comfort and a help," Sebastian replied, "but we must not
give it you to-night. Live as you have been living, in the love of our Lord
and in obedience to Father Antonio. For the present you can wear no habit
more acceptable to God than the coat in which you do your daily duty about
the farm. Do not hang your head. I foresee that an abbot will once more rule
within these walls, and that you, José will die as one of his family. Have
patience."

A sudden change came over Sebastian as he ceased speaking. The hectic


bloom faded from his cheeks, and the heavy lids drooped over his
preternaturally bright eyes. A moment later he sank forward against the
table. Antonio and José sprang at once to his help. He had swooned. They
made haste to bear him bodily to his cell. It was an easy task; for beyond
the weight of his cloak there seemed to be hardly anything to carry. After
they had laid him on his bed and dashed water from the torrent in his face,
he revived and said faintly:
"Thanks, thanks, thanks, I am well. Leave me. I shall say mass to-
morrow at five o'clock. Leave me."

He fell into another unnatural sleep. But Antonio did not leave him. All
through the short warm night he watched and prayed. At last the dull chant
of the Atlantic was drowned under the glittering trills of near blackbirds.
Day dawned. The sun rose above Sebastian's Spain; and the sleeper awoke.

He answered the traditional Benedicamus Domino with so ringing a Deo


gratias that Antonio thought a miracle had happened. Sebastian looked
stronger and healthier than ever before. Even José, who had been sleeping
heavily on the corridor floor, was aroused by Sebastian's two words.

They repaired to the chapel. There Father Sebastian heard the


confessions of his two companions. Without delay he proceeded to the
sacristy. Antonio followed him and began to lift from its drawer one of the
less costly vestments which had never been taken away. It was green and
gold, as appointed in the Ordo for that day. But Sebastian, having bidden
him replace it, drew forth a black chasuble, simply embroidered with a
plain white cross. Antonio felt justly rebuked. When the Abbot was dead,
and the Prior and all the fathers save two, surely it was meet that the
survivor's Mass should be a Mass of requiem.

From his pocket-case, Sebastian took the unconsecrated wafers which


he had brought from Lisbon. He finished his vesting and preparation and
they re-entered the chapel. José was kneeling devoutly on the lowest step of
the sanctuary. Outside, hundreds of birds were in full boisterous song.

Father Sebastian went to the foot of the altar and began to say Mass. He
uttered the words quickly and clearly, and made the genuflections without
difficulty. Indeed, Antonio, as he poured water over the white and fleshless
fingers at the psalm Lavabo, marveled more than ever at the miracle of his
friend's sudden strength. At the commemoration of the dead, the intensity of
Sebastian's recollection seemed to make the whole chapel thrill and throb,
like a bed of reeds in a wind.

After he had given the most holy Body to Antonio and to José,
Sebastian concluded the Mass and returned to the sacristy with a firm tread.
He laid aside the sacred vestments and came back to his old stall in order to
make his thanksgiving. Antonio, also in his old stall, knelt at Sebastian's
side.

The ascending sun cleared the top of the hill and shone into the chapel.
The diadem of the Holy Child blazed with glory. In all the trees happy birds
redoubled their songs.

Half an hour passed. José, arising noisily, made Antonio open his eyes.
But Father Sebastian knelt without moving against the sloping book-board.
José clattered out. Still Father Sebastian did not move. Antonio waited,
revering his friend's ecstasy of communion with his Lord. He waited long.
But meanwhile a broad sunbeam had been working westward; and at last it
poured its burning gold upon the bended head.

Antonio was stepping softly forward to screen his friend from the fierce
ray when a sudden instinct bade him kneel down and look into Sebastian's
face. But Sebastian's wide-open, rapturous eyes did not gaze into Antonio's;
nor were they beholding any earthly thing. So beautiful was the sight that
Antonio's exclamation was more a shout of joy than a cry of fear. Into his
mind there rushed the words of Isaias which had been Sebastian's favorite
scripture in the old days, Regem in decore suo videbunt oculi ejus: "His
eyes shall see the King in His beauty; they shall behold the land which is
very far off."

Antonio and José buried the body of Sebastian that night on the sunny
side of the cloister, between the third and fourth pillars, just under the tile-
picture of Enos, with its legend, Ambulavit cum Deo et non apparuit, quia
tulit eum Deus: "He walked with God and was no more seen, for God took
him."
IV

On the feast of Saints Peter and Paul, just thirty days after Sebastian's
death, Antonio heard Mass in the village church. Forty-eight hours were left
to him before his payment to the Villa Branca Fazenda became due. In the
strong-box at home he had only three hundred and twelve pounds towards
his debt of five hundred. Nothing had been received from Sebastian's friend
in Spain, although sufficient time had elapsed for a reply to reach the farm.
Nevertheless, Antonio rose from his knees at the end of Mass and took his
way homeward with a serene spirit.

From the point where he and José had seen the ruts of young
Crowberry's wheels nearly two years before, the monk heard thumping
hoofs. He gazed down the road and saw an advancing cloud of dust. A few
moments later he made out the milk-white Branco which had succeeded
coal-black Negro as the Navares' post-horse. Thomé, the postman, drew
rein and handed Antonio two letters.

The first was from young Crowberry. It ran:

Dear Friend da Rocha.

You will be sorry to hear that my father died last week, suddenly. I know
you will pray for him; and I hope you will pray for me too.

Strange to say, Sir Percy also passed away last week, two days after my
father. I saw it in the papers, but I know no details. At Christmas my father
saw him at Weymouth, and he seemed well.

As our affairs are tangled, I have much to do. Write to me soon. My


thoughts turn to you very often nowadays. Tell me how you do, all round. I
remain, your sincere friend. Edward Crowberry.

You might also like