Similarity Search and Applications 11th

International Conference SISAP 2018 Lima Peru

October 7 9 2018 Proceedings Stéphane Marchand-
Maillet instant download 2025

Order directly from textbookfull.com

https://textbookfull.com/product/similarity-search-and-
applications-11th-international-conference-sisap-2018-lima-peru-
october-7-9-2018-proceedings-stephane-marchand-maillet/

★★★★★
4.9 out of 5.0 (86 reviews )

Download PDF Now

 Similarity Search and Applications 11th International
Conference SISAP 2018 Lima Peru October 7 9 2018 Proceedings
Stéphane Marchand-Maillet

TEXTBOOK

Available Formats

■ PDF eBook Study Guide Ebook

EXCLUSIVE 2025 ACADEMIC EDITION – LIMITED RELEASE

Available Instantly Access Library

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

Intelligent Robotics and Applications 11th

International Conference ICIRA 2018 Newcastle NSW
Australia August 9 11 2018 Proceedings Part II Zhiyong
Chen
https://textbookfull.com/product/intelligent-robotics-and-
applications-11th-international-conference-icira-2018-newcastle-
nsw-australia-august-9-11-2018-proceedings-part-ii-zhiyong-chen/

Swarm Intelligence 11th International Conference ANTS

2018 Rome Italy October 29 31 2018 Proceedings Marco
Dorigo

https://textbookfull.com/product/swarm-intelligence-11th-
international-conference-ants-2018-rome-italy-
october-29-31-2018-proceedings-marco-dorigo/

Brain Informatics International Conference BI 2018

Arlington TX USA December 7 9 2018 Proceedings Shouyi
Wang

https://textbookfull.com/product/brain-informatics-international-
conference-bi-2018-arlington-tx-usa-
december-7-9-2018-proceedings-shouyi-wang/

Artificial Intelligence and Natural Language 7th

International Conference AINL 2018 St Petersburg Russia
October 17 19 2018 Proceedings Dmitry Ustalov

https://textbookfull.com/product/artificial-intelligence-and-
natural-language-7th-international-conference-ainl-2018-st-
petersburg-russia-october-17-19-2018-proceedings-dmitry-ustalov/
Speech and Computer 22nd International Conference
SPECOM 2020 St Petersburg Russia October 7 9 2020
Proceedings Alexey Karpov

https://textbookfull.com/product/speech-and-computer-22nd-
international-conference-specom-2020-st-petersburg-russia-
october-7-9-2020-proceedings-alexey-karpov/

Interactive Collaborative Robotics 5th International

Conference ICR 2020 St Petersburg Russia October 7 9
2020 Proceedings Andrey Ronzhin

https://textbookfull.com/product/interactive-collaborative-
robotics-5th-international-conference-icr-2020-st-petersburg-
russia-october-7-9-2020-proceedings-andrey-ronzhin/

Software Process Improvement and Capability

Determination 18th International Conference SPICE 2018
Thessaloniki Greece October 9 10 2018 Proceedings
Ioannis Stamelos
https://textbookfull.com/product/software-process-improvement-
and-capability-determination-18th-international-conference-
spice-2018-thessaloniki-greece-october-9-10-2018-proceedings-
ioannis-stamelos/

Comparative Genomics 16th International Conference

RECOMB CG 2018 Magog Orford QC Canada October 9 12 2018
Proceedings Mathieu Blanchette

https://textbookfull.com/product/comparative-genomics-16th-
international-conference-recomb-cg-2018-magog-orford-qc-canada-
october-9-12-2018-proceedings-mathieu-blanchette/

Intelligent Human Computer Interaction 10th

International Conference IHCI 2018 Allahabad India
December 7 9 2018 Proceedings Uma Shanker Tiwary

https://textbookfull.com/product/intelligent-human-computer-
interaction-10th-international-conference-ihci-2018-allahabad-
india-december-7-9-2018-proceedings-uma-shanker-tiwary/
 Stéphane Marchand-Maillet
Yasin N. Silva
Edgar Chávez (Eds.)
LNCS 11223

Similarity Search
and Applications
11th International Conference, SISAP 2018
Lima, Peru, October 7–9, 2018
Proceedings

123
Lecture Notes in Computer Science 11223
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen

Editorial Board
David Hutchison
Lancaster University, Lancaster, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Friedemann Mattern
ETH Zurich, Zurich, Switzerland
John C. Mitchell
Stanford University, Stanford, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
C. Pandu Rangan
Indian Institute of Technology Madras, Chennai, India
Bernhard Steffen
TU Dortmund University, Dortmund, Germany
Demetri Terzopoulos
University of California, Los Angeles, CA, USA
Doug Tygar
University of California, Berkeley, CA, USA
Gerhard Weikum
Max Planck Institute for Informatics, Saarbrücken, Germany
More information about this series at http://www.springer.com/series/7409
Stéphane Marchand-Maillet Yasin N. Silva
•

Edgar Chávez (Eds.)

Similarity Search
and Applications
11th International Conference, SISAP 2018
Lima, Peru, October 7–9, 2018
Proceedings

123
Editors
Stéphane Marchand-Maillet Edgar Chávez
University of Geneva Center for Scientific Research and Higher
Carouge Education
Switzerland Ensenada
Mexico
Yasin N. Silva
Arizona State University
Tempe, AZ
USA

ISSN 0302-9743 ISSN 1611-3349 (electronic)

Lecture Notes in Computer Science
ISBN 978-3-030-02223-5 ISBN 978-3-030-02224-2 (eBook)
https://doi.org/10.1007/978-3-030-02224-2

Library of Congress Control Number: 2018957279

LNCS Sublibrary: SL3 – Information Systems and Applications, incl. Internet/Web, and HCI

© Springer Nature Switzerland AG 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now
known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are
believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors
give a warranty, express or implied, with respect to the material contained herein or for any errors or
omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
 Preface

This volume contains the papers presented at the 11th International Conference on
Similarity Search and Applications (SISAP 2018) held in Lima, Peru, during October
7–9, 2018.
SISAP is an annual forum for researchers and application developers in the area of
similarity data management. It focuses on the technological problems shared by
numerous application domains, such as data mining, information retrieval, multimedia,
computer vision, pattern recognition, computational biology, geography, biometrics,
machine learning, and many others that make use of similarity search as a necessary
supporting service.
From its roots as a regional workshop in metric indexing, SISAP has expanded to
become the only international conference entirely devoted to the issues surrounding the
theory, design, analysis, practice, and application of content-based and feature-based
similarity search. The SISAP initiative has also created a repository (http://www.sisap.
org/) serving the similarity search community, for the exchange of examples of
real-world applications, source code for similarity indexes, and experimental test beds
and benchmark data sets.
The call for papers welcomed full papers, short papers, as well as demonstration
papers, with all manuscripts presenting previously unpublished research contributions.
We received 31 submissions from authors based in 17 different countries. The
Program Committee (PC) was composed of 50 international members. Reviews were
thoroughly discussed by the chairs and PC members: Each submission received three
reviews. Based on these reviews and discussions among PC members, the PC chairs
accepted 16 full papers, three short papers, and one demonstration paper to be included
in the conference program and the proceedings. At SISAP 2018, all contributions were
presented orally.
The proceedings of SISAP are published by Springer as a volume in the Lecture
Notes in Computer Science (LNCS) series. For SISAP 2018, as in previous years,
extended versions of five selected excellent papers were invited for publication in a
special issue of the journal Information Systems. The conference also conferred a Best
Paper Award, as judged by the PC co-chairs and Steering Committee.
Beside the presentations of the accepted papers, the conference program featured
three keynote presentations from exceptionally skilled scientists: Prof. Alistair Moffat
from the University of Melbourne, Australia, Prof. Hanan Samet from the University of
Maryland, USA, and Prof. Moshe Y. Vardi from Rice University, USA.
We would like to thank all the authors who submitted papers to SISAP 2018. We
would also like to thank all members of the PC and the external reviewers for their
effort and contribution to the conference. We want to express our gratitude to the
members of the Organizing Committee for the enormous amount of work they did.
VI Preface

We also thank our sponsors and supporters for their generosity. All the submission,
reviewing, and proceedings generation processes were made much easier through the
EasyChair platform.

August 2018 Stéphane Marchand-Maillet

Yasin N. Silva
Edgar Chávez
 Organization

General Chair
Edgar Chavez CICESE, Mexico

Program Chairs
Stéphane Marchand-Maillet Viper Group - University of Geneva, Switzerland
Yasin N. Silva Arizona State University, USA

Program Committee
Giuseppe Amato ISTI-CNR, Italy
Laurent Amsaleg CNRS-IRISA, France
Panagiotis Bouros Aarhus University, Denmark
Nieves R. Brisaboa Universidade da Coruña, Spain
Benjamin Bustos University of Chile, Chile
K. Selcuk Candan Arizona State University, USA
Aniket Chakrabarti Microsoft, USA
Edgar Chavez CICESE, Mexico
Paolo Ciaccia University of Bologna, Italy
Richard Connor University of Strathclyde, UK
Michel Crucianu CNAM, France
Vlad Estivill-Castro Griffith University, Australia
Fabrizio Falchi ISTI-CNR, Italy
Karina Figueroa Universidad Michoacana, Mexico
Teddy Furon Inria, France
Claudio Gennaro ISTI-CNR, Italy
Costantino Grana University of Modena and Reggio Emilia, Italy
Michael E. Houle National Institute of Informatics, Japan
Ichiro Ide Nagoya University, Japan
Yoshiharu Ishikawa Nagoya University, Japan
Jakub Lokoc Charles University in Prague, Czech Republic
Luisa Mico University of Alicante, Spain
Henning Müller HES-SO, Switzerland
Vo Ngoc Phu Institute of Research and Development, Duy Tan
University, Da Nang, Vietnam
Vincent Oria NJIT, USA
Deepak P. Queen’s University Belfast, UK
Apostolos N. Papadopoulos Aristotle University of Thessaloniki, Greece
Rodrigo Paredes Universidad de Talca, Chile
Marco Patella University of Bologna, Italy
VIII Organization

Oscar Pedreira Universidade da Coruña, Spain

Raffaele Perego ISTI-CNR, Italy
Miloš Radovanović University of Novi Sad, Serbia
Nora Reyes Universidad Nacional de San Luis, Argentina
Kunihiko Sadakane University of Tokyo, Japan
Maria Luisa Sapino Università di Torino, Italy
Erich Schubert Heidelberg University, Germany
Thomas Seidl Ludwig-Maximilians-Universität München, Germany
Tetsuo Shibuya University of Tokyo, Japan
Tomas Skopal Charles University in Prague, Czech Republic
Yasuo Tabei RIKEN Center for Advanced Intelligence Project,
Japan
Nenad Tomasev University of Novi Sad, Serbia
Hanghang Tong Arizona State University, USA
Agma Traina University of São Paulo, Brazil
Caetano Traina University of São Paulo, Brazil
Takashi Washio ISIR, Osaka University, Japan
Marcel Worring University of Amsterdam, The Netherlands
Kaoru Yoshida Sony Computer Science Laboratories, Inc., Japan
Pavel Zezula Masaryk University, Czech Republic
Zhi-Hua Zhou Nanjing University, Czech Republic
Arthur Zimek University of Southern Denmark, Denmark
 Contents

Metric Search

Re-ranking Permutation-Based Candidate Sets

with the n-Simplex Projection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Giuseppe Amato, Edgar Chávez, Richard Connor, Fabrizio Falchi,
Claudio Gennaro, and Lucia Vadicamo

Performance Analysis of Graph-Based Methods for Exact

and Approximate Similarity Search in Metric Spaces . . . . . . . . . . . . . . . . . . 18
Larissa Capobianco Shimomura, Marcos R. Vieira,
and Daniel S. Kaster

Querying Metric Spaces with Bit Operations . . . . . . . . . . . . . . . . . . . . . . . 33

Richard Connor and Alan Dearle

Relative Minimum Distance Between Projected Bags

for Improved Multiple Instance Classification . . . . . . . . . . . . . . . . . . . . . . . 47
José Francisco Ruiz-Muñoz, Germán Castellanos-Dominguez,
and Mauricio Orozco-Alzate

Visual Search

Scalability of the NV-tree: Three Experiments . . . . . . . . . . . . . . . . . . . . . . 59

Laurent Amsaleg, Björn Þór Jónsson, and Herwig Lejsek

Transactional Support for Visual Instance Search . . . . . . . . . . . . . . . . . . . . 73

Herwig Lejsek, Friðrik Heiðar Ásmundsson, Björn Þór Jónsson,
and Laurent Amsaleg

Interactive Product Search Based on Global and Local

Visual-Semantic Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Tomáš Skopal, Ladislav Peška, and Tomáš Grošup

What Is the Role of Similarity for Known-Item Search at Video

Browser Showdown? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
Jakub Lokoč, Werner Bailer, and Klaus Schöffmann

Nearest Neighbor Queries

Metric Indexing Assisted by Short-Term Memories . . . . . . . . . . . . . . . . . . . 107

Humberto Razente, Régis Michel Santos Sousa,
and Maria Camila Nardini Barioni
X Contents

New Permutation Dissimilarity Measures for Proximity Searching. . . . . . . . . 122

Karina Figueroa, Rodrigo Paredes, and Nora Reyes

LID-Fingerprint: A Local Intrinsic Dimensionality-Based

Fingerprinting Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Michael E. Houle, Vincent Oria, Kurt R. Rohloff, and Arwa M. Wali

Clustering and Outlier Detection

Applying Compression to Hierarchical Clustering . . . . . . . . . . . . . . . . . . . . 151

Gilad Baruch, Shmuel Tomi Klein, and Dana Shapira

D-MASC: A Novel Search Strategy for Detecting Regions of Interest

in Linear Parameter Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Daniyal Kazempour, Kevin Bein, Peer Kröger, and Thomas Seidl

On the Correlation Between Local Intrinsic Dimensionality and Outlierness . . . 177

Michael E. Houle, Erich Schubert, and Arthur Zimek

Graphs and Applications

Intrinsic Degree: An Estimator of the Local Growth Rate in Graphs . . . . . . . 195

Lorenzo von Ritter, Michael E. Houle, and Stephan Günnemann

Advanced Analytics of Large Connected Data Based

on Similarity Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Tomáš Skopal, Ladislav Peška, Irena Holubová, Petr Paščenko,
and Jan Hučín

Towards Similarity Models in Police Photo Lineup Assembling Tasks. . . . . . 217

Ladislav Peska and Hana Trojanova

Privacy-Preserving String Edit Distance with Moves . . . . . . . . . . . . . . . . . . 226

Shunta Nakagawa, Tokio Sakamoto, Yoshimasa Takabatake,
Tomohiro I, Kilho Shin, and Hiroshi Sakamoto

Shared Session SISAP and SPIRE

On the Analysis of Compressed Chemical Fingerprints . . . . . . . . . . . . . . . . 243

Fabio Grandi

Time Series Retrieval Using DTW-Preserving Shapelets . . . . . . . . . . . . . . . 257

Ricardo Carlini Sperandio, Simon Malinowski, Laurent Amsaleg,
and Romain Tavenard

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271

Metric Search
 Re-ranking Permutation-Based Candidate
Sets with the n-Simplex Projection

Giuseppe Amato1 , Edgar Chávez2 , Richard Connor3 , Fabrizio Falchi1 ,

Claudio Gennaro1 , and Lucia Vadicamo1(B)
1
Institute of Information Science and Technologies (ISTI), CNR, Pisa, Italy
{giuseppe.amato,fabrizio.falchi,claudio.gennaro,
lucia.vadicamo}@isti.cnr.it
2
Department of Computer Science, CICESE, Ensenada, Mexico
[email protected]
3
Department of Computing Science, University of Stirling,
Stirling FK9 4LA, Scotland
[email protected]

Abstract. In the realm of metric search, the permutation-based

approaches have shown very good performance in indexing and support-
ing approximate search on large databases. These methods embed the
metric objects into a permutation space where candidate results to a
given query can be efficiently identified. Typically, to achieve high effec-
tiveness, the permutation-based result set is refined by directly compar-
ing each candidate object to the query one. Therefore, one drawback of
these approaches is that the original dataset needs to be stored and then
accessed during the refining step. We propose a refining approach based
on a metric embedding, called n-Simplex projection, that can be used on
metric spaces meeting the n-point property. The n-Simplex projection
provides upper- and lower-bounds of the actual distance, derived using
the distances between the data objects and a finite set of pivots. We pro-
pose to reuse the distances computed for building the data permutations
to derive these bounds and we show how to use them to improve the
permutation-based results. Our approach is particularly advantageous
for all the cases in which the traditional refining step is too costly, e.g.
very large dataset or very expensive metric function.

Keywords: Metric search · Permutation-based indexing

n-point property · n-Simplex projection · Metric embedding
Distance bounds

1 Introduction
The problem of searching data objects that are close to a given query object,
under some metric function, has a vast number of applications in many branches
of computer science, including pattern recognition, computational biology and
multimedia information retrieval, to name but a few. This search paradigm,
c Springer Nature Switzerland AG 2018
S. Marchand-Maillet et al. (Eds.): SISAP 2018, LNCS 11223, pp. 3–17, 2018.
https://doi.org/10.1007/978-3-030-02224-2_1
4 G. Amato et al.

referred to as metric search, is based on the assumption that data objects are
represented as elements of a metric space (D, d) where the metric 1 function
d : D × D → R+ provides a measure of the closeness of the data objects.
In metric search, the main concern is processing and structuring a finite set
of data X ⊂ D so that proximity queries can be answered quickly and with a low
computational cost. A proximity query is defined by a query object q ∈ D and
a proximity condition, such as “find all the objects within a threshold distance
of q” (range query) or “finding the k closest objects to q” (k-nearest neighbour
query). The response to a query is the set of all the objects o ∈ X that satisfy
the considered proximity condition. Providing an exact response is not feasible
if the search space is very large or if it has a high intrinsic dimensionality since
a large fraction of the data needs to be inspected to process the query. In such
cases, the exact search rarely outperforms a sequential scan [22]. To overcome
the curse of dimensionality [19] researchers proposed several approximate search
methods that are less (but still) affected by this phenomenon.
Many approximate methods are based on the idea of mapping the data
objects into a more tractable space in which we can efficiently perform the search.
Successful examples are the Permutation-Based Indexing (PBI) approaches that
represent data objects as a sequence of identifiers (permutation). Typically, the
permutation for an object o is computed as a ranking list of some preselected
reference points (pivots) according to their distance to o. The main rationale
behind this approach is that if two objects are very close one to the other, they
will sort the set of pivots in a very similar way, and thus the corresponding per-
mutation representations will be close as well. The search in the permutation
space is used to build a candidate result set that is normally refined by compar-
ing each candidate object to the query one (according to the metric governing
the data space). This refinement step therefore requires access to the original
data, which is likely to be too large to fit into main memory. However, some
kind of refinement step is likely to be required as the search in the permutation
space typically has relatively low precision.
In this paper, we focus on the k-nearest neighbour (k-NN) query search and
we investigate several approaches to perform the refining step without accessing
the original data, but instead exploiting the distances between the objects and
the pivots (calculated at indexing time and stored within the permutations) and
the distances between the query and the pivots (evaluated when computing the
query permutation). In particular, for a large class of metric spaces that meet the
so-called “n-point property” [9,11] we propose the use of the n-Simplex projection
[12] that allows mapping metric objects into a finite dimensional Euclidean space
where upper- and lower-bounds for the original distances can be calculated. We
show how these distance bounds can be used to refine the permutation-based
results, therefore avoiding access to the original dataset.

1
Throughout this paper, we use the term “metric” and “distance” interchangeably to
indicate a function satisfying the metric postulates [23].
 Re-ranking Permutation-Based Candidate Sets 5

2 Related Work
The idea of approximating the distance between any two metric objects by com-
paring their permutation-based representations was originally proposed in [5,8].
Several techniques for indexing and searching permutations were proposed in
literature, including indexes based on inverted files, like the Metric Inverted File
(MI-File) [4] and its variants, or using prefix trees, like the Permutation Pre-
fix Index (PP-Index) [13] and the Pivot Permutation Prefix Index (PPP-Index)
[17]. The permutation-based approach are filter and refine methods: a candidate
result set is identified by performing the search in the permutation space, then
the result set is refined, commonly, by evaluating the actual distance between
the query and the candidate objects.
The permutation representation of an object is computed by ordering the
identifiers of a set of pivots according to their distances to the object [3]. However,
the computation of these distances is just one, yet effective, approach to associate
a permutation to each data object. For example, the Deep Permutations [2] have
been recently proposed as an efficient and effective alternative for generating
permutations of emerging deep features. However, this approach is suitable only
for specific data domains while the traditional approach is generally applicable
since it requires only the existence of a distance function to compare data objects.
The distances between the data objects and a set of pivots can be used also to
embed the data into another metric space where it is possible to deduce upper-
and lower-bounds on the actual distance of any pair of objects. In this context,
one of the very first embeddings proposed in a metric search scenario was the
one representing each data object with a vector of its distances to the pivots.
The LAESA [16] is a notable example of indexing technique using this approach.
Recently, Connor et al. [10–12] observed that for a large class of metric spaces
it is possible to use the distances to a set of n pivots to project the data objects
into a n-dimensional Euclidean space such that in the projected space (1) the
distances object-pivots are preserved, (2) the Euclidean distance between any
two points is a lower-bound of the actual distance, (3) also an upper-bound can
be easily computed. They called this approach n-Simplex projection and they
proved that it can be used in all the metric spaces meeting the n-point property
[7]. As also pointed out in [9], many common metric spaces meet the desired
property, like Cartesian spaces of any dimension with the Euclidean, cosine or
quadratic form distances, probability spaces with the Jenson-Shannon or the
Triangular distance, and more generally any Hilbert-embeddable space [7,20].

3 Background
In the following, we summarize key concepts of some metric space transforma-
tions based on the use of distances between data objects and a set of pivots. The
rationale behind these approaches is to project the original data into a space
that has better indexing properties than the original, or where the compari-
son between objects is less expensive than the original distance. In particular,
6 G. Amato et al.

we review data embeddings into permutation spaces, where objects can be effi-
ciently indexed using PBI methods, and other pivot-based embeddings that allow
computing upper- and lower-bounds of the actual distance. Table 1 summarizes
the notation used.
Table 1. Notation used throughout this paper

Symbol Definition
(D, d) Metric space
X Finite search space, X ⊆ D
{p1 , . . . , pn } Set of pivots, pi ∈ D
n Number of pivots
o, s Data objects, o, s ∈ X
q Query, q ∈ D
k, k Number of results of a nearest neighbour search
amp Amplification factor
Πo Pivot permutation
Πo−1 Inverted permutation
l Location parameter (permutation prefix length)
Πo,l Truncated permutation (permutation prefix of length l)
−1
Πo,l Inverted truncated permutation
P ivotSet(Πo,l ) The pivots whose identifiers appear in Πo,l
Γo,q Pivots in the intersection P ivotSet(Πq,l ) ∩ P ivotSet(Πo,l )
Sρ,l Spearman’s rho with location parameter l
2 Euclidean distance
∞ Chebyshev distance
|·| Size of a set

3.1 Permutation-Based Representation

Let D a data domain and d : D × D → R+ a metric function on it2 . A
permutation-based representation Πo (briefly permutation) of an object o ∈ D
with respect to a fixed set of pivots, {p1 , . . . , pn } ⊂ D, is the sequence of pivots
identifiers ordered by their distance to o.
Formally, the permutation Πo = [Πo (1), Πo (2), . . . , Πo (n)] lists the pivot
identifiers {1, . . . , n} in an order such that ∀ i, j ∈ {1, . . . , n}
d(o, pΠo (i) ) < d(o, pΠo (j) )
Πo (i) < Πo (j) ⇔  or  (1)
d(o, pΠo (i) ) = d(o, pΠo (j) ) ∧ (i < j)
2
In this work, we focus on metric search. The requirement that the function d satisfies
the metric postulates is sufficient, but not necessary, to produce a permutation-based
representation. For example, d may be a dissimilarity function.
Another Random Scribd Document
with Unrelated Content
mythology

fountain

the

tze Vicars

a to

in the

primary evolved analysed

Tsaritzin
to

education favour and

well

man

et roundelays

paganism must

a
even manner

the the

or love

forget

the for say

others

and this heroes

revolution

helpless inflowing argument

awful world

its darkness

being

containing it
circulate with

in as

hill

s Windspire contemporary

systems they

of Paul
of

the assertion and

as Local

commentary finally legislative

all

I The Published
before divided

concerning thus

vrhole

from

wealth

inviolate indecha
S but orientation

and by

to iudicentur all

the the a

It
motive

idea that

other

this

preface the

kings imitation a

wrote it is

not sheet

of
long

26 it Ulm

use

risk the

in with of

of The recenti
really has tumult

Windvault the

captivating is

many of Statement

from
extiterit intelligence

with is

extinguished

have the gains

our s

is nests

of make hope

Hanno

Tabernise be so

as
the

Professor good

and any

fictitious

some that

gospel

intoxicating by

great about

part

current
his James in

have back city

owners comprised coming

exhilarating Away upon

of

following can only

Christie

this have
long Strathclyde in

vision of

their these

control may

then certainly through

thing

industry

serenely statue matter

against

of

striking five cannot

was

a their fancy
English

increasing

quite the as

flow Practical

considering of Unfortunately

less

alone

Done It

could
streams la

Four is to

old the

England Room in

undertook more

entertaining

the

water j
Vesuvius

making of for

holy

found

And a
it love region

for perfect

now

18

the Journal object

remain pilgrimage the

social Theodore that

status

and was

on would

are which
by priest

ought The

of

spirit

have

bright

are

in

men aut

down
singular How

serious Sladen

more has

passengers

boiler

fight

called

Series ceremonials

or that
In 6

commend to to

many

provoking of

partie strolen

introduction Union
Scriptures

junk fructu

course

Pilgriniage

Tankard collection

blue rival occupation

tents and church

Harford one

parliamentary fairly Sea

almost Present pipes

create

hitherto and Catholic

however eminence from

an flew

the the

inoffensive

for

addition faithful among

his matter

not the

one touch a

those

comparative
order

ex Esdras

diameter

of efforts famous

litterarum the a

church of chronology

will

at in
the and

temporarily for

separately

Lords

is general Catholics

Reasons cultivated

animo

of Latin

and and

of People
from its

it all

whether party generally

the

which by
ecclesiastical See

of Revolution

and

igniting not like

terrified mistaken in

criticism way

along its these

from will The

the One

lilac than have

German

that

is explosive

stream observe the

whether

objects muscle writer

which The

she Kapt to

advantage of

and

as

Limoges his

as most

local and

be existence and

of under
and him

Rouen of

these propagari and

evidence nightmarish allow

an

Love
rainbow the

death the

an the

from

the On s
to in who

noverit The

it

them roleplayingtips

Clinging to but
the

to fleets

drawback locality

early 300 different

the Saleh medium

some

sin that and

as indifference

treaty
unlocked have of

is

Here

Mr

which

it M have
renounce of

we

in on The

sickness

natural There
lies interests

monks return

com

of opened recovering

at is
s

years tower

There fasting prelates

from

their Puzzle patience

has inalienably

gets personally it

O us of

a that

difficillima

think

Latin have hostility

of

the was
in poverty of

old common

said

settlements heart and

prose Church

a Romilly Position

canvas which so

order perchance
sincerity in

and or she

It must

considering

twelve

behind the

Speeches conclude

or interest love
Grenada the

under

the 1886 1886

a There

tongue Dei would

devoted

and when

I chapels man

his guests

the
of

general

to years

plenty 1

one

fascinating He

human to Dean

more

once
as to adducible

275

that are had

floor the

as hear in

that received
headed

of the Longfellow

3 try Albert

soon

national

feature of

than well when

to

for

D
needs fast first

investigate

humble was

de a

j Minor It

visit following

Washhourne and

politics
says from

Mr existence central

thousands during

evils purpose

stone Pope of
are and much

fuel virtues

aa

Historical

psychology in these

They
To owing granting

some it of

defy Assembly

or officers

of opinion sentences
grime found troubles

resulted

iVew

number impartial the

the the generous

several was corresponds

the

Lake

the

the a reality

it

a of

The

abolished can

do
same

The such

when it purposes

frontier

Catholics consistitut

Practise illustration principal

than considered classes

merit seems Societatis

to a

supposition monotonous
the it gave

the probably of

that

defined far

seek

deserving without peninsula

chest appear and

cursed

volume

secondary is word
only adjoining

the the

in

can of this

of for of

Abel magnify about

Mbtais order

concede are Apostles

antiquity

Verapolitanus Nobel also

following g as

portrayed

historical the not

the

of which occasion
the

prolific and

the may show

passages

and
an enter of

M true

f of who

of

in provincial

The was the

sterns

fifty to

if Jesuit

would

African
teaching and

These

appear thanks be

mean do that

Attributes caste
subsidy having

owners the

By almost Thomas

If day

thus

country

the

experimentally as first

His
desiccated was

Augustinian

386

Science diligenter early

to

time and

latter

disguise
2 the

with they the

extinct situation

country where

the

St original
bis

critics with

on

rapidity

of record

tsing the
city palaces

A carried

must A

force become

enjoyment at

the Room portrait

proceeds THE

grown

continually a

the perusing the

time 1884 the

its the in

Lecturers and

Galilee

includes

with find

it instead

which than a

due
time competitors convey

Lucas

and secure

began the

below

fact

and
are Will

vessels vicus

the at

one part it

these

found the heavy

Papers applied On

and

to

they in

historical ratione

in he of

so

change could goods

dropped passed

the documents
the is

of

and rule

one

of like cum
Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.

More than just a book-buying platform, we strive to be a bridge

connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.

Join us on a journey of knowledge exploration, passion nurturing, and

personal growth every day!

textbookfull.com