ESWC 2007

From LeibnizWiki
Jump to: navigation, search

Contents

ESWC Day 1

Invited Lecture - Stefano Ceri

  • Power of modeling -> dealing with web innovation.

WebML makes the distinction between the (orthogonal) aspects:

  • Data
  • Hypertext -> logic of web interfacing
  • Presentation

Implemented in an eclipse environment: WebRatio (used to build e.g. the Acer company website)

  • Hypertext: Provides siteview, area, page, unit, operation. The design process is a loop... this loop can be traversed in 5 minutes
  • Extend WebML to deal with innovation --> add process modeling steps to the methodology.
    • Generate skeleton WebML from BPM using XSLT
    • Define a standard sub-model for processes
  • Extend with web-services
    • From communication with users to communication with services
    • Provide a standard sub-model of services
    • Units: WSDL support & XML transformations
  • Semantic Services using WSMO -> how to obtain _automatic_ semantic annotation of services?
    • Extra steps (using WSMO Flight)
      • Import ontologies
      • Semantic Annotation
    • Extra units
      • Query units to graphically build queries
    • Generate WSML ontology from MebML (capability, choreography and goal generator)

Session: Case Studies

IMPORTNet - Mechatronics Domain - Damjanovic

  • Mechatronics involves a lot of collaboration between engineers (different sub-fields)
  • There is a lot of correspondence between ontology and mechatronic engineering
  • Did a survey of existing methodology:
    • Methontology, Diligent, On-To-Knowledge, are all 'agnostic' wrt. the knowledge representation language
    • Therefore chose DynamONT, which is not agnostic but has DOLCE as its target ontology representation language.
    • Uses DOLCE's ontology design patterns
  • Did not use pre-existing ontologies except DOLCE
  • Did not give a clear answer to the question why they use DOLCE, apparently the DynamONT is the 'official' methodology in Austria.

Media, Politics and the Semantic Web - Wouter van Atteveld

  • Interplay between media, politics, public.
  • Want to measure _media content_
    • Automatic text analysis (difficult) vs manual text analysis (expensive)
    • Need for large datasets. Important to share, combine, reuse.
    • Interoperability of *primitives* -> standardization
    • Network Translation Analysis (methodoloty in social science), actually quite cool.

SEEMP - E. Della Valle

  • Employment Mediation Marketplace
    • Minimal shared commitment to communicate between vocabularies of multiple _public_ employment services
  • Uses WSMT (Webservice Modelling Toolkit), an Eclipse application.

ExpertFinder - Malgorzata Mogol and Axl

  • Reuses and combines FOAF, and other vocabularies to describe experts, their knowledge, and interconnections.

See http://rdfweb.org for ExpertFinder

Invited Lecture - Ning Zhong

About Web Intelligence:

  • Applying techniques of IT&AI to the web
  • Introduces new challenges
  • Enhancement of AI & IT


Session

"Empowering Software Maintainers" - Rene Witte et al.

  • Ontologies to check & generate instances form java code and software documentation
  • Use the ontology to do code analysis.

Any World Access to OWL from Prolog - Pascal Hitzler and Tobias Matzner

Implemented "Any world semantics" (Karlsruhe) using SWI-Prolog and KAON2

PrOWLog: http://logic.aifb.uni-karlsruhe.de/wiki/PrOWLog

Posters

SemDAV

Semantic 'file system'-like interface. Looks really nice, at http://www.semdav.org

ESWC Day 2

Invited Lecture - Gottlob

Implemented the LiXto system. An ELog system, that supports a monadic datalog for traversing webpages and do data extraction. There's a wvisual wrapper, you can create queries/templates through a GUI. It's free too.

Session

Extract RDF from Wikipedia Infoboxes - Auer

http://aksw.org and http://dbpedia.org (also on sourceforge).

Quite nice, as it allows you to query a lot of information implicit in Wikipedia.

  • Many "literals" remain after translation, i.e. this makes it rather slow
  • Supports extraxtion but no feed-back into Wikipedia
  • .... SNORQL (?)
  • Applicable to all wikipedia installations
  • Future: query wikipedia directly through a SPARQL end-point

Semantic Annotation of Scientific Publications - Tudor Groza and Siegfried Handschuh

SALT.

Supports LaTeX, but should also be usable for other formats.

  • Generates XMP (PDF metadata) trhough the use of a LaTeX package.
  • Allows for inline N3 annotations
  • Text structure based on standards such as RST and ABCDE

See http://salt.semanticweb.org

Mining the web through verbs - Inescid, Lisboa

Heuristics -> relating entities to the closest verbs.

Cross-domain, shallow heuristics.

Invited Lecture - Ron Brachman

Important article about Web 2.x by Mark Greaves in IEEE Intelligent Systems

New "categories" of computer science: bring together traditional disciplines:

    • "Finding"-science & systems.
    • Community science & systems
    • Algorithmic advertising
    • Computational Micro-economics (computer science & economics)
    • Media Experiences
    • Data science
      • Nature, volume & flow in large companies
      • Data as a macro phenomenon

Finding Science

  • From IR to contextual information supply.
  • Social search: the people factor
  • It's about results, not documents
    • 40-65% of searches is 'to learn something', informational
    • 25-25% of searches is navigational
    • 35-20% of searches is transactional (service access, download, shop)
  • Issues: representation of context, user and structure

Community Science

  • Social networks: need sociologists
  • Understanding online communities
  • "Six Degrees", the interconnected age (book)
  • Four types of communities: social, enthousiasts, knowledge, marketplace
  • Search within a group? What is the right group?
  • Example: the DBLife portal for the db community supports social search
  • Example: http://answers.yahoo.com
  • Facebook
  • Yahoo "myweb 2.0" augment with community search
  • "The wisdom of crowds", Janis Surowiecki (book) (markets)

Remember:

  • Ads pay the bills
  • Incentive systems
  • Match Ads to user

(see the Web Science Initiative at WWW)

Session

TeleTask

Semantic composition of lecture subparts. Uses DL (not planning). This is somewhat similar to process descriptions.

Personal tools