Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Web 3.0 Emerging


Published on

A talk given at CSHALS 2010, an overview of the current state of linked data, Semantic Web, and Web 3.0 efforts.

Published in: Education

Web 3.0 Emerging

  1. 1. Web 3.0 emerges… Jim Hendler Tetherless World Professor of Computer and Cognitive Science Assistant Dean of Information Technology and Web Science Rensselaer Polytechnic Institute
  2. 3. The Semantic Web (ca. 2001) (Berners-Lee, Hendler, Lassila; 2001)
  3. 4. Semantic Web ca. 2010 <ul><li>Semantic Web finding success even in tough market </li></ul><ul><ul><li>Lots of small companies in the market: Altova… Zepheira (eg. C&P, Franz, Intellidimension, Intellisophic, Ontology Works, Siderean, SandPiper, SiberLogic, TopQuadrant …) </li></ul></ul><ul><ul><li>Web 3.0 new buzzword: Garlik, Twine, Freebase, Bintro, Siri, Talis, … </li></ul></ul><ul><ul><li>Semantic Search taking off - Powerset bought by Microsoft for over $100,000,000, hakia, bing, T2, tiptop, … </li></ul></ul><ul><ul><li>Semantic match: classifieds (bintro), clinical studies ( </li></ul></ul><ul><li>Bigger players buying in </li></ul><ul><ul><li>2009 announcements at SemTech (June): Google, New York Times, Oracle, IBM, Yahoo, MS Live Labs, Siri, … </li></ul></ul><ul><ul><li>2008: Gartner identifies Corporate Semantic Web as one of three &quot;High impact&quot; Web technologies </li></ul></ul><ul><ul><li>Tool market forming: AllegroGraph, TopBraid, Pellet2, … </li></ul></ul><ul><ul><li>O’Reilly “Programming the Semantic Web” </li></ul></ul><ul><li>Government projects in and across agencies </li></ul><ul><ul><li>Recent open data announcements by UK and US </li></ul></ul><ul><ul><ul><li>UK in linked data format, US 3 rd party to linked data (5B triples so far) </li></ul></ul></ul><ul><ul><li>Projects/demos in EU, Japan, Korea, China, India… </li></ul></ul><ul><ul><li>SKOS update in govt (and private) libraries </li></ul></ul><ul><li>Several &quot;verticals&quot; heavily using Semantic Web technologies </li></ul><ul><ul><li>Health Care and Life Sciences </li></ul></ul><ul><ul><li>Financial services </li></ul></ul><ul><ul><li>Human Resources </li></ul></ul><ul><ul><li>Publishing/New Media </li></ul></ul><ul><ul><li>Sciences other than Life Science </li></ul></ul><ul><ul><ul><li>Virtual observatory, Geo ontology, … </li></ul></ul></ul>
  4. 5. Two very different sorts of use cases <ul><li>cf. US National Center for Biotechnology Information, &quot;Oncology Metathesaurus&quot; </li></ul><ul><ul><li>50,000+ classes, ~8 people supporting full time, monthly updates, mandated for use by NIH-funded cancer researchers </li></ul></ul><ul><ul><ul><li>OWL DL rigorously followed </li></ul></ul></ul><ul><ul><ul><li>Provably consistent </li></ul></ul></ul><ul><li>cf. Friend of a Friend (Foaf) </li></ul><ul><ul><li>30+ classes, Dan Brickley and Libby Miller made it, maintained by consensus in a small community of developers </li></ul></ul><ul><ul><ul><li>Violates DL rules (undecidable) </li></ul></ul></ul><ul><ul><ul><li>Used inconsistently </li></ul></ul></ul>
  5. 6. Widely varying use <ul><li>NCBI Oncology Ontology </li></ul><ul><ul><li>High use in medical community </li></ul></ul><ul><ul><li>Very &quot;trusted&quot; information (provenance from NCBI) </li></ul></ul><ul><ul><li>Primarily terminological (relationships between cancer-related concepts), not data-oriented </li></ul></ul><ul><li>FOAF </li></ul><ul><ul><li>~60M Foaf people (not necessarily distinct individuals) </li></ul></ul><ul><ul><li>Used by a number of large providers </li></ul></ul><ul><ul><ul><li>If you use LiveJournal, you have a FOAF file </li></ul></ul></ul><ul><ul><ul><ul><li>Also flickr, ecademy, tribe, joost, … </li></ul></ul></ul></ul><ul><ul><ul><ul><li>And you can export Foaf from Facebook and many other social networking sites </li></ul></ul></ul></ul><ul><ul><li>Becoming de facto standard for open social networking </li></ul></ul>
  6. 7. Why ? <ul><li>NCBI view: Formal properties </li></ul><ul><ul><li>Based on a decidable subset of KR </li></ul></ul><ul><ul><ul><li>Description logics </li></ul></ul></ul><ul><ul><li>For which much scaling research has been happening </li></ul></ul><ul><ul><ul><li>Ca. 2000 - 10,000 axioms, no facts, 1 day </li></ul></ul></ul><ul><ul><ul><li>Ca. 2008 - 50,000 axioms, million facts, 10 min. </li></ul></ul></ul><ul><ul><ul><ul><li>Not just faster computers (but Moore's Law helps), significant research into optimization, &quot;average case&quot; </li></ul></ul></ul></ul><ul><ul><ul><ul><li>Moving to parallel (Web server) </li></ul></ul></ul></ul><ul><ul><li>With some new ways of linking to larger data sets </li></ul></ul><ul><ul><ul><li>SHER, IBM, &quot;reduced Abox&quot; </li></ul></ul></ul><ul><ul><ul><li>OWL-Prime, Oracle, &quot;materialized views” </li></ul></ul></ul><ul><ul><ul><li>OWL 2 QL (?) </li></ul></ul></ul>In this view OWL is a formal knowledge representation standard
  7. 8. The argument for this seems compelling <ul><li>When &quot;folksonomy&quot; isn't enough… </li></ul>Which one do you want your doctor to use?
  8. 9. But the cost is high <ul><li>Formal modeling finds its use cases in verticals and enterprises </li></ul><ul><ul><li>Where the vocabulary can be controlled </li></ul></ul><ul><ul><li>Where finding things in the data is important </li></ul></ul><ul><li>Example </li></ul><ul><ul><li>Drug discovery from data </li></ul></ul><ul><ul><ul><li>Model the molecule (site, chemical properties, etc) as faithfully and expressively as possible </li></ul></ul></ul><ul><ul><ul><li>Use &quot;Realization&quot; to categorize data assets against the ontology </li></ul></ul></ul><ul><ul><ul><ul><li>Bad or missed answers are money down the drain </li></ul></ul></ul></ul><ul><li>But the modeling is very expensive and the return on investment must be very high! </li></ul><ul><ul><li>Which is part of why the &quot;expert systems revolution&quot; wasn't one </li></ul></ul><ul><ul><li>Became part of the technology tool kit, a useful niche in the programming pantheon, but didn't change the world </li></ul></ul>
  9. 10. The alternative <ul><li>OWL is based on RDF, a language designed for the (Semantic) Web </li></ul><ul><ul><li>Built with Web architecture in mind </li></ul></ul><ul><ul><ul><li>Exploits Web infrastructure, respects W3C TAG recommendations </li></ul></ul></ul><ul><ul><ul><ul><li>Internationalization, accessibility, extensibility </li></ul></ul></ul></ul><ul><ul><li>Fits the Web culture </li></ul></ul><ul><ul><ul><li>Open and extensible, supports communities of interest </li></ul></ul></ul><ul><ul><ul><ul><li>If you don't like my ontology, extend it, change it, or build your own </li></ul></ul></ul></ul><ul><ul><ul><li>Fits the Web application development paradigm </li></ul></ul></ul><ul><ul><ul><ul><li>Scales like &quot;databases&quot; </li></ul></ul></ul></ul><ul><ul><li>With some new ways of linking to formal models </li></ul></ul><ul><ul><ul><li>Heavy use of a small amount of OWL </li></ul></ul></ul><ul><ul><ul><li>Generally used &quot;like it sounds&quot; not like the formal model </li></ul></ul></ul><ul><ul><ul><ul><li>Example &quot;owl:sameAs&quot; debate </li></ul></ul></ul></ul>“ linked data” often used to describe this low semantics Semantic Web (slogan: a little semantics goes a long way)
  10. 11. Semantic Web Applications <ul><li>~2006: Web app developers discover the Semantic Web </li></ul>RDF Triple Store Dynamic Content Engine HTTP RDF Web App (w SPARQL) RDF Triple Store … HTML 2008 examples include sites from &quot;regular&quot; Web players such as Dow Jones, Reuters and Yahoo!
  11. 12. cf. Yahoo mixes RDF with other technologies: at Web scale Dave Beckett, SemTech 08
  12. 13. Linked Data + Semantics <ul><li>&quot;Linked Data&quot; approach finds its use cases in Web Applications (at Web scales) </li></ul><ul><ul><li>A lot of data, a little semantics </li></ul></ul><ul><ul><li>Finding anything in the mess can be a win! </li></ul></ul><ul><li>Example </li></ul><ul><ul><li>Declare simple inferable relationships and apply, at scale, to large, heterogeneous data collections </li></ul></ul><ul><ul><ul><li>eg. Use InverseFunctional triangulation to find the entities that can be inferred to be the same </li></ul></ul></ul><ul><ul><ul><ul><li>These are &quot;heuristics&quot; not every answer must be right (qua Google) </li></ul></ul></ul></ul><ul><ul><ul><ul><li>But remember time = money ! </li></ul></ul></ul></ul>
  13. 14. The data is out there Example: the linked open data cloud now has tens of billions of triples and is growing rapidly
  14. 15. Government Data on the Web
  15. 16. Moving to linked data (UK) <ul><li>Built around linked data with top-down push from “Number 10” </li></ul>
  16. 17. Moving to linked data (US) <ul><li>Third parties (like RPI) translate the govt data into Sem Web forms and link to sources </li></ul>• Plans for a in OGD implementation plans,, but unfunded
  17. 18. Adding Meta-data
  18. 19. Pump through to Google Viz for demos
  19. 20. +
  20. 21. Adding some Web magic Web Analytics Social Data Networks External Links
  21. 22. Mashup w/Web content
  22. 23. Linked Data (RDF, SPARQL) Semantic Web (RDFS, owl) Web 3.0 Web 2.0 Web 3.0 extends current Web applications using Semantic Web technologies and graph-based, open data.
  23. 24. Semantic Search IEEE Computer, Jan 2010; IEEE Computing Now, Feb 2010 (free)
  24. 25. Semantic Search examples T2 ( TipTop ( )
  25. 26. Web 3.0 examples Semantic classified (
  26. 27.
  27. 28. Web 3.0 examples
  28. 29. Web 3.0 examples Social database (
  29. 30. Tiptop health Ok, so we have a way to go on this one 
  30. 31. Web 3.0 excitement (hype?) <ul><li>Significant and growing commercial interest… </li></ul><ul><ul><li>Web: Google, Amazon, Travelocity… </li></ul></ul><ul><ul><li>Web 2.0: Facebook, Wikipedia, YouTube, Twitter… </li></ul></ul><ul><ul><li>Web 3.0: the big ones are still out there </li></ul></ul>
  31. 32. 1B computers (mostly owned by a few large companies) 3B cell phones (most are Web enable)
  32. 33. Sem Web going mobile? (Add social contexts)
  33. 34. Summary <ul><li>The Semantic Web is going just fine thank you </li></ul><ul><ul><li>People asking “how,” not why </li></ul></ul><ul><li>So far the commercial driver has been “weak semantics” </li></ul><ul><ul><li>In the enterprise </li></ul></ul><ul><li>Web 3.0 adds semantics as a value add to regular Web functionality </li></ul><ul><ul><li>Semantic search </li></ul></ul><ul><ul><li>Semantic match </li></ul></ul><ul><ul><li>Semantic etc </li></ul></ul><ul><li>The big one is still out there </li></ul>
  34. 35. What’s next? Still, like many researchers working in this area, Hendler is already looking beyond emerging Semantic Web strategies and related technologies that are now collectively called Web 3.0. “This stuff is new and exciting,” he says. “ But I look at it this way: I started playing with the Semantic Web back in the 1990s. As a researcher, I’m not content to sit around and exploit Web 3.0; my job is to help create Web 4.0.” “Engineering the Web’s Third Decade,” CACM March, 2010