1 / 25

Semantic Knowledge Management Finding Information Through Meaning Not Words

Semantic Knowledge Management Finding Information Through Meaning Not Words . Alistair Duke alistair.duke@bt.com Next Generation Web Research. The Semantic Web Is Dead. Mor Naaman Yahoo! Research Berkeley Panellist at WWW2007, Banff.

omer
Download Presentation

Semantic Knowledge Management Finding Information Through Meaning Not Words

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Semantic Knowledge ManagementFinding Information Through Meaning Not Words Alistair Duke alistair.duke@bt.com Next Generation Web Research

  2. The Semantic Web Is Dead Mor Naaman Yahoo! Research Berkeley Panellist at WWW2007, Banff “the grand vision of 'A Semantic Web' will not be achieved, mostly because users cannot be expected to annotate media with complex labels … but can only be expected to use simple tags”

  3. The need for semantics • Knowledge workers overwhelmed by info • from intranets, emails, newslines … • but still lack vital information • 80% of corporate data is unstructured • including key business decisions • subject to regulation, e.g. SOX • Companies suffer from • decisions made under incomplete knowledge • threat of compliance failure

  4. We need information … • Identified by semantics, not just keywords • precise and complete • Selected by their interests & task context • defined semantically • From heterogeneous sources, • accessed uniformly • Presented meaningfully • and appropriately for the user

  5. Semantic Information Management: In three words Semantic information management classifies, finds, distributes, shares and uses information based on meaning not on the particular words used to represent meaning.

  6. In three words Semantic information management classifies, finds, distributes, shares and uses information based on meaningnot on the particular words used to represent meaning.

  7. The SEKT Project • Addressing the semantic knowledge technology research agenda • European 6th framework IP project • end date 31/12/2006 • 36 months duration, €12.5m budget • www.sekt-project.com

  8. The inSEKTs Vrije Universiteit Amsterdam Siemens Empolis University of Sheffield Universität Karlsruhe BT Ontoprise Kea-pro Universität Innsbruck iSOCO Sirma AI Universitat Autònoma de Barcelona Jozef Stefan Institute

  9. Major research challenges • Improve automation of ontology and metadata generation • Research and develop techniques for ontology management and evolution • Develop highly-scalable solutions • Research sound inferencing despite inconsistent models • Develop semantic knowledge access tools • Develop methodology for deployment

  10. Extracting the semantics • Information extraction • using human language technology • Knowledge discovery • machine learning and statistical methods • Existing metadata, e.g. database schemas • mapping and merging

  11. Semantic Annotation

  12. Precision in Semantic Web Search • Semantic Search could match • a query: Documents concerning a telecom company inEurope with John Smith as a director • With a document containing: “At its meeting on the 10th of May, the board of Vodafone appointed John Smith as CTO" • Traditional search engines cannot do the required reasoning: • Vodafone is a mobile operator, which is a kind of telecom company; • Vodafone is in the UK, which is a part of Europe; • CTO is a type of director

  13. PROTON – the SEKT Ontology • PROTON - a light-weight upper-level ontology; • 250 NE classes; • 100 relations and attributes; • covers mostly NE classes, general concepts and KM concepts • Mappings to DC, FOAF, RSS, DOLCE http://proton.semanticweb.org/

  14. PROTON World KB • PROTON is populated with a “world knowledge base” • Aims to cover the most popular entities in the world • Collected from various sources, like geographical and business intelligence gazetteers. • Organizations: business, international, political, government, sport, academic… • Specific people, (e.g. politicians) • Locations: countries, regions, cities, etc. • Automatic identification of these entities within documents indexed • 2m+ OWL statements

  15. KAON2 Reasoner: Mapping to relational model KAON2 Reasoner: Rules Engine hasCoAuthor Author hasWritten Publication KAON2 Rules Mapping Author id name …. hasWritten autherId publicationId …. Publication id title ….

  16. Search and Browse in SEKT SEKTagent “A Semantic Search Alerting Service” Squirrel “A Semantic Search and Browse Tool”

  17. SEKTagent Overview • PROTON-based Semantic queries • Periodic alerts of matching results • Highlights queried entities in results (also related entities) • Natural Language summaries of ontological knowledge • Device Independence • PC, Palm and Mobile

  18. SektAgent Demonstration

  19. Squirrel - Overview • Hybrid approach - combines free text and semantic search • Ontology based browsing • Meta-result to help guide search • Use of rules and reasoning through KAON2 • Natural Language summaries of ontological knowledge • User profile based result ranking

  20. Squirrel Demonstration

  21. Result consolidation:Delivering summaries to users instead of a list of links • Identifying the most relevant parts of documents returned as query responses • Results presented as consolidated summaries. • Reduces the need for users to navigate to and read multiple documents • Document segments and their relevance are determined via • analysis of the frequency of named entities in the text • proximity of the text to the user's query and interest profile. • Semantically Enhanced ‘Text-Tiling’

  22. Query: ‘Hurricane Katrina’

  23. Text-tiling using named entities • Hurricane Katrina is thought to have killed hundreds, probably thousands of people in New Orleans, the city's mayor, Ray Nagin, has said. Mr Nagin said there were significant numbers of corpses in the waters of the flood-stricken city, while many more people may be dead in their homes. • There would be a total evacuation of the city, he said, warning it could be months before residents could return. • President George W Bush said the area could take years to recover. • Cutting short a holiday in Texas to take charge of the federal recovery effort, Mr Bush said the government was dealing with one of the worst natural disasters in US history. • "This is going to be a difficult road, the challenges we face on the ground are unprecedented, but there's no doubt in my mind that we'll succeed," he said. • Mr Bush, whose Air Force One plane flew low over the affected area, was taken aback by the scale of the disaster. Classification against topic ontology Politics US Local Government US Federal Government

  24. Result Consolidation Demonstration

  25. For more information http://www.keapro.net/sekt

More Related