"Why the Semantic Web will Never Work" (note the quotes)
Download as PPTX, PDF32 likes33,187 views
The document discusses the challenges and successes of the semantic web, arguing that while it has made significant strides, it has not fully achieved its original vision. It critiques common misconceptions and addresses the need for improved integration between linked-data and ontology views, emphasizing the importance of research to overcome existing barriers. The conclusion asserts that the semantic web is functioning but requires updates and unification of its models to reach its full potential.
"Why the Semantic Web will Never Work" (note the quotes)
- 1. “Why the Semantic Web will Never Work”(note the quote marks!)Jim HendlerRPIhttp://www.cs.rpi.edu/~hendler@jahendler(sorry, not in rhyme)
- 2. Friends, Romans (& Greeks), CountrymenLend me your earsI have come to bury the semantic web, not to praise it
- 3. Does it meanWhy our critics were wrong when they said “The Semantic Web will Never Work” orWhy the Semantic Web will Never Achieve the Vision we had for it (at least if we don’t fix things)
- 4. Yes (not Xor)OutlineSome current Semantic Web SuccessesRevisit the Semantic Web visionWhat did we say we would doReview successes and failuresWhat has worked as well (or better) than we expectedWhat hasn’tWhat are some challenges to overcome to achieve the latter?
- 6. History>200 Semantic Web talks since 2000
- 7. Pre-HistoryWho first conceived of the Semantic Web?Tim Berners-Lee (WWW Geneva, 1994)"This is a pity, as in fact documents on the web describe real objects and imaginary concepts, and give particular relationships between them... For example, a document might describe a person. The title document to a house describes a house and also the ownership relation with a person. ... This means that machines, as well as people operating on the web of information, can do real things. For example, a program could search for a house and negotiate transfer of ownership of the house to a new owner. The land registry guarantees that the title actually represents reality.”Tim Berners-Lee plenary presentation at WWW Geneva, 1994
- 8. Beyond XML:Agent SemanticsPrehistory: 1st funding talk Oct. 1999DARPA will lead the way with the development of Agent markup Language (DAML)a “semantic” language that ties the information on a page to machine readable semantics (ontology)Currently being explored at University levelSHOE (Maryland), Ontobroker(Karlsruhe),OWL(Washington Univ)Largely grows from past DARPA programs (I3, ARPI)But not transitioning W3C focused on short-term gain:HTML/XML<ONTOLOGY ID=”powerpoint-ontology" VERSION="1.0" DESCRIPTION=”formal model for powerpoint presentations"><DEF-CATEGORY NAME=”Title" ISA=”Pres-Feature" > <DEF-CATEGORY NAME=”Subtitle" ISA=”Pres-Feature" ><DEF-RELATION NAME=”title-of" SHORT="was written by"> <DEF-ARG POS=1 TYPE=”presentation"> <DEF-ARG POS=2 TYPE=”presenter" ><Title> Beyond XML <subtitle> agent semantics </subtitle> </title><USE-ONTOLOGY ID=”PPT-ontology" VERSION="1.0" PREFIX=”PP" URL= "http://iwp.darpa.mil/ppt..html"><CATEGORY NAME=”pp.presentation” FOR="http://iwp.darpa.mil/jhendler/agents.html"> <RELATION-VALUE POS1 = “Agents” POS2 = “/jhendler”>
- 9. Berners-Lee et al, 2001(May 21, 2001)
- 10. usesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesusesThis leads to a radically new view of interoperationDistributed,partially mapped, inconsistent -- but very flexible!
- 11. But, like the web…
- 12. DAMLNotional ScheduleNowLater2001: We will change the world!
- 13. Web “travel agents”How many cows are there in Texas?Query processed: 73 answers foundGoogle document search finds 235,312 possible page hits.Http://www…/CowTexas.html claims the answer is 289,921,836A database entitled “Texas Cattle Association” can be queried for the answer, but you will need “authorization as a state employee.”A computer program that can compute that number is offered by the State of Texas Cattleman’s Cooperative, click here to run program....The “sex network” can answer anything that troubles you, click here for relief... The “UFO network” claims the “all cows in Texas have been replaced by aliens“Agent” Markup Language
- 15. Animal ontology
- 16. Use that markup in query/portal interfaces
- 17. Services need Web Logics2001: Semantic Web Services
- 18. Services off the desktop2003: Semantic Web Services
- 19. So where have we got toSemantic Web technology use has exceeded even my wildest expectationsWhat is different now?Semantic SearchAll the big kids are playing!Advertising drives Web markets“Markets are created by disaggregating the producer and the consumer” “Buzz” around data on the Web esp. Open Government Data
- 20. Example: OGP use growing quicklyFacebook incentivizing use of RDFa like buttons15,178 sites of top 1,000,000 as of 3/3/11Oct 2010: FB reportsRDFa is ~ 10-15% of > 3,000,000 likes per day!Facebook is encouraging developers to use the RDFaversion
- 21. Because they want the links!The network is where their money is made! (predicted >$5B of advertising in next two years)
- 22. Creates a platform for SW-powered apps
- 23. They said it couldn’t be doneCommon Criticisms
- 24. The Shirky fallacy Folksonomy will winTagging the technology of choiceTagging has largely failed to meet its promise
- 25. Tagging doesn’t achieve goals without “social context”
- 26. Example: Flickr tag “James”; Amazon tag “My-…”The Network effect requires links (Hendler & Golbeck, JWS, 2008)
- 27. The database community fallacyThe semantic web will never scale,1,000,000 triples and things go to heck Winner of the 2009 Billion Triples ChallengeJust plain wrong!!
- 28. “ad hoc” data integrationexample: Linked Open Govt DataMore than 50 of these at http://logd.tw.rpi.eduSee also http://data.gov and http://data.gov.uk
- 29. And we do things the DB community struggles with
- 30. Another Shirky criticismThis is just a make-work program to keep AI scientists busy doing what they’ve always doneCannot create an ontology at Web ScaleAI never works so it won’t this timeLogic and reasoning will not work on the Web because people disagree and because logic isn’t powerful enough for what is needed(ok, he called it syllogism, but we know what he meant)
- 31. Sem Web 2010April 2010
- 32. Semantic Web 2010July 2010
- 33. Sem Web 2010August 2010
- 34. Sem Web 2010July 2010
- 35. Sem Web 2010August 2010
- 37. The “bottom” of the Semantic WebWhat isseeing the mostuse??RDFa
- 38. The success of “Linked Data”Maturation of RDF technologiesSPARQL endpointsFits Web development modelsRDFaWorks well with current search paradigmsA little semantics goes a long wayBUT WHAT IS STUNNING IS JUST HOW LITTLE!Equality via same URIRDFa mostly w/DBMS not triple storeNot only no reasoning, but hardly any “principled” inferencing!
- 39. The bad news…The ontology story is still confused
- 40. Decidable Logic basisinconsistencyOntology: the OWL DL viewOntology as Barad-Dur (Sauron's tower):Extremely powerful!Patrolled by OrcsLet one little hobbit in, and the whole thing could come crashing down
- 41. ontology: the linked-data viewontology and the tower of BabelWe will build a tower to reach the skyWe only need a little ontological agreementWho cares if we all speak different languages?Genesis 11:7 Let us go down, and there confound their language, that they may not understand one another's speech. So the Lord scattered them abroad from thence upon the face of all the earth: and they left off to build the city.
- 42. OWL has had successesExamples from Clark and Parsia (2011)Decision-support tool for sales people to automate policy driven cross-selling recommendations at very large US bank built out of RDF integrated data, OWL reasoning, and PelletAt global 25 company (another bank) OWL and Pellet form the core of a bank-wide Entitlements service to represent, analyze, and query every access control policy for the entire bank, globally, in 50+ legal jurisdictionsAnd many other companies could claim similarBut most of these sorts of systems are still just coming out of prototype phaseAnd most are still more “expert” system than Web app
- 43. The tough love stuffOWL is succeeding to a large degree as a KR standardBuilding “expert systems” as a business has never gone away; OWL improves toolingBut it is largely failing in bringing representation to the WWWcf. “misuse” of owl:sameAs >> “proper” usecf. rdf:class >> owl:classcf. it is rare that ontologies link to others
- 44. The gap is growingLinked-Data-based applications are growing in size, number and importance on the WebBut the “vocabulary” story is still unclearOntology research is turning OWL into a usable KR standard,But the linking story is still unclearNo linking without vocabulariesNo network effect without links
- 45. What I think we MUST doBridging the gap between the linked-data and ontology views requires some key research challenges to be addressedDL (and FOL) are useful formalisms for KR&R, but do not address the needs of the Web!Empirical comparisons are useful in scaling systems, but do not address the needs of an academic community!
- 46. My Challenge to youA sufficient formalism for Semantic Web applications mustProvide a model that accounts for linked data What is the equivalent of a DB calculus?Provide a means for evaluating incomplete reasonersIn practice we must be able to model A-box effects as formally as T-box technologies
- 47. Be bold!A sufficient formalism for Semantic Web applications must alsoDefine what an ontology isontologies really areIncluding external referents linking between termsIncluding ontology alignment partial mapping Including non-expressive formalisms real-world “errors”
- 48. It just might work…One idea on how to get thereDefine common problems that offer features of interest to both communitiesCompare approaches with respect to performanceDevelop hybrids that have best features of both as necessaryRepeat(thanks Bettina!)
- 49. SummaryIADIS-2008The infrastructure needs of intelligent systems are now being met by a combination of Semantic Web, Linked Data, Web Services and Rule-based systemsKnowledge engineering can be jumpstarted from existing terminologies/ontologies, semi-structured systems, and other Web resourcesWeb Services (espWSDL, SAWSDL) provide "wrappers" and other methods to let "legacy" systems play with agentsReasoners and rule-based systems are scaling in new ways, and receiving some standardizationSo where are all the agents???
- 50. Conclusion: “Why the Semantic Web will never work”?No reason at all The Semantic Web is here, it is working, and it will continue to do soBut, for it to move to the next level and be all that we as a community have aspired forWe must revisit and update the early visions for the modern webWe must unify the “competing” models of linked-data and machine-readable vocabulariesWe must step up to some critical research challenges
- 51. AppendixResearch Challenges (ca. 2008)
- 52. Research ChallengesWhat is the Web culture?Design/use/analysis are connected to "cultural stereotypes" (Think HSBC ads)What are the cultural stereotypes in the emerging online community?What level of "knowledge" is needed by Web users? Is this dependent on application? User community? Is expressivity a plus, minus, non-issue?Especially in an open system (previous AI systems were "closed"
- 53. Research ChallengesComputational challenges as "end user" supportScalingSemantic Web HCI (What do we show "real users"?)What are the trade-offs in useVirtually all AI literature assumes a high-cost, high-value modelThe Semantic Web is showing us alternative models What are the trade-offs, analysesIf more and more of what we see includes integrated data from multiple sources, will that change the trust modelsDo we need to expose provenance? Will "provider" model be changed?
- 54. Research ChallengesWho are the "experts"What level of expertise is needed to become "dangerous" with this new technology?What is the "ecosystem" (what is the equivalent of Web developer/web master/web user?) If more and more of what we see includes integrated data from multiple sources, will that change the trust modelsDo we need to expose provenance? Will "provider" model be changed? Formal vs. informal models of ontologyI didn't discuss "folksonomy" but a key aspect is "social context" (Hendler & Golbeck, 08)Can social contexts use
- 56. Questions?