Rethinking Microblogging: Open Distributed Semantic

Digital Enterprise Research Institute www.deri.ie

Rethinking Microblogging:
Open, Distributed, Semantic
Alexandre Passant, John G. Breslin, Stefan Decker

Digital Enterprise Research Institute, NUI Galway
http://deri.ie

ICWE2010
Thursday, 8th July 2010
Vienna, Austria
© Copyright 2009 Digital Enterprise Research Institute. All rights reserved.

Agenda

  Microblogging: current state and issues
  Requirements to enhance microblogging
  SMOB – Semantic MicroBLogging
  An ontology stack for Social Semantic Web applications
  Distributed hubs and communication protocols
  Integration with the Linking Open Data (LOD) cloud
  Browsing, discovering and querying
  Integrated capabilities (end-user interface, SPARQL, maps)
  External indexes and components
  Future Work
  Conclusion

Microblogging

  Short status update, generally < 140 chars
  Real-time information management / Citizen-sensing
  Popular on the Web (Twitter) and in the enterprise (Yammer)

Limits of current systems

  And of Web 2.0 systems in general
  Walled-garden systems / Lack of portability
  Data belongs to the service used to publish it
  Lack of structure / semantics
–  Recent work on OpenGraph and Twitter Annotations
  « A Bill of Rights for Users of the Social Web »

  Most research on microblogging focuses on
communication patterns and social behaviours
  But what about engineering issues ?

Requirements

  Machine-readable metadata (R1)
  Making microblogging systems more interoperable
  Focus on microblog posts, content and authors

  Distributed architecture and open data (R2)
  Solving the walled garden and data portability issues
  Letting users own and control their data

  Data reuse and interlinking (R3)
  Interlinking microblog posts with other initiatives
  Reusing existing data to make more sense of microblogging

SMOB – Semantic MicrOBlogging

  Our proposal
  A framework for open, distributed and semantic
microblogging
  Based on state-of-the-art Semantic Web technologies (RDF(S)/
OWL, RDFa, SPARQL) and Linked Data principles to provide
the previous requirements

  SMOB - http://smob.me:
  Open-source framework (GPL)
  Started mid-2008, completely re-designed end-2009
–  Distributed architecture, LOD-integration, etc.

The Semantic Web and Linked Data

  Semantic Web
  From documents to structured data
  Annotations (RDF), ontologies (RDFS/OWL), queries (SPARQL)
  Linked Data
  A set of principles for publishing data on the Web
  Linking Open Data project – interlinking datasets on the Web
using the LD principles

SMOB and our initial requirements

  Machine-readable metadata (R1)
  An ontology stack for microblogging
  Representation of posts in RDFa, SPARQL endpoint

  Distributed architecture and open data (R2)
  Distributed hubs spread over the Web
  Interacting via HTTP + SPARQL/Update

  Data reuse and interlinking (R3)
  Interlinking microblog (and their posts) with other systems
  Reusing existing data when available

Ontologies for microblogging (R1)

  Different needs
  Users and Profiles (Personal information and Social Networks)
  Presence (Geolocation, current activity, etc.)
  Data (microblog containers and microblog posts)
  Topics (#tags)

  Our approach
  Integrating and extending existing lightweight ontologies to
focus on modularity and reusability of components
  Providing a complete ontology stack for Semantic
Microblogging, and more broadly for any Social Semantic
Web application

FOAF – People and Social Networks

  FOAF – Friend Of A Friend
  http://foaf-project.org
  An ontology to describe people and their relationships
  Can be integrated with any other SW vocabularies
  FOAF on the Web
  hi5, LiveJournal, Drupal 7, etc. and exporters for popular
services

FOAF – Distributed user-profiles

  Ability to reuse existing profiles
  Such as RDFa-enabled documents (e.g. Drupal 7)
  No need to duplicate personal data

foaf:depiction in RDFa
http://apassant.net

http://example.org

SIOC – Profiles and data

  SIOC – Semantically-Interlinked Online Communities
  http://sioc-project.org
  Representing online communities and their content
  W3C Member Submission
  A types module for finer-grained content-types

SIOC – Profiles and data

  SIOC extensions
  Current state of SIOC cannot caputure all the properties of
microblogging

  New Classes
  sioct:MicroBlog : Microblog container
  sioct:MicroblogPost : Microblog post

  New properties
  sioc:follows : following / followers (directed graph model)
  sioc:addressed_to : @reply patterns

OPO – Presence Information

  OPO – Online Presence Ontology
  http://online-presence.net
  Representing rich presence information using semantics
  Geolocation, current activity (project, etc.) …
  Integration with SIOC to map content to the one’s presence

MOAT – Semantic Tagging

  MOAT – Meaning Of A Tag
  http://moat-project.org
  A model to provide semantic tagging capabilities
  Linking #tags to their meanings (defined as URIs)
  Provides integration with the Linking Open Data cloud
Tag Ontology
tag:RestrictedTagging

rdf:type http://example.org/tag/
apple
tag:associatedTag
http://example.org/
tagging1

tag:taggedBy tag:taggedResource

moat:tagMeaning

http://example.org/ http://dbpedia.org/
moat:taggedWith resource/Apple_Inc.
post/1

foaf:maker dct:title

Nouvel iPhone rdf:type MOAT + DBpedia
http://apassant.net/alex/
disponible

sioct:BlogPost
FOAF
SIOC + DC

The SMOB Ontology Stack

  Integration of the previous components
  Plus a smob:Hub class to represent user’s hubs
  Can be reused in various Social Semantic Web contexts
  Makes Social Web applications part of the LOD cloud

Representation

  Each microblog post is represented in RDF using the
aforementionned ontology stack
  Dereferenceable URI for each post
  Subset directly in XHTML pages using RDFa (/page)
  Complete representation also available using Turtle (/data)

Distributed architecture (R2)

  Personal SMOB hubs spread all over the Web
  No central server / no dependency to 3rd-party services
  Ensure data ownership and privacy
  Each hub simply requires LAMP settings (based on ARC2)
and provides a SPARQL endpoint
  Can be used as read-write Twitter clients

  Following / follower registration
  Ability to get « remote followers », represented internally
(in both the follower and followee hub) in RDF
  :user_a sioc:follows :user_b .!

Communication between hubs

  Replication / notification between peers
  Broadcasting data to followers when new content is created
  Using SPARQL/Update via HTTP POST (Checking access rights)
  Simple HTTP POST to Twitter API

Integration with the LOD cloud (R3)

  Semantic tagging
  URIs being suggested on runtime when typing #tags
  Integration of microblogging within the LOD cloud
(Dbpedia lookup, Sindice)
  Plug-in system to add new service (e.g. enterprise KB)

Geolocation

  Geolocation features
  Run-time geolocation lookup using GeoNames.org
  Modelled as part of the user’s presence

Interlinking benefits

  Benefits of LOD interlinking
  Can re-use backgound knowledge when querying data
–  E.g. Microblog posts about any city in Italy (will retrieve #Torino)
  Microblog content becomes more discoverable

Browsing, discovering and querying

  End-user interface
  Genuine microblogging interface, on top of RDF data
  Map view
  Geolocation capabilities provided thanks to GeoNames
  Integration with Sindice
  Third-party service for identifying SMOB hubs and content
  SPARQL endpoint
  Direct queries and pluggable components via HTTP

End-user interface

  Genuine microblogging interface
  Generated using SPARQL queries
  Integration of Twitter posts (also stored in RDF)
  RDFa markup for each post and user

Map view

  Geolocation features
  Benefits of the GeoNames lookup integration
  Reusing coordinates provided by the GeoNames KB (in RDF)

Integration with Sindice

  Sindice – the Semantic Web index
  http://sindice.com
  SMOB hubs can ping Sindice when new content is created
  Retrieving distributed SMOB data from a single entry point
  Transversal SPARQL querying to discover microblog posts

SPARQL endpoint

  Each hub provides its own endpoint
  Using SPARQL, no need to learn a new API
  Direct queries sent via HTTP / answers as JSON/XML
  Ability to plug-in external components (e.g. Explorator)

Future Work

  Scalability
  PubSubHubbub integration (cf. recent work on sparqlPuSH)
  Cache system for rendering SPARQL-based pages
  Modelling
  Full-RDFa / content-negociation
  Interlinking and data generation
  Nanosyntaxes (generating RDF data about updates’ content)

  Uptake
  Spread the word to enrich the number of SMOB hubs
deployed on the Web !

Conclusion

  Contributions
  An Ontology stack for Social Semantic Web applications
  A distributed architecture for microblogging
  Integration of microblogging with the LOD cloud
  Deployed in SMOB – http://smob.me

  Take-home message
  There are opportunities for a distributed Social Semantic Web
  SMOB can be just a part of this ecosystem, it’s up to you !
  Semantic Web and Linked Data provide straightworward
integration of other components following the same approach

Thank you !

  http://smob.me
  GNU/GPL, runs on any LAMP environment
  Regular new features and bugfixes, consider the SVN version

  Main contact
  http://apassant.net
  alexandre.passant@deri.org
  @terraces

Rethinking Microblogging: Open Distributed Semantic

More Related Content

What's hot (20)

Viewers also liked (15)

Similar to Rethinking Microblogging: Open Distributed Semantic (20)

More from Alexandre Passant (20)

Recently uploaded (20)

Rethinking Microblogging: Open Distributed Semantic