1. Making the Web Searchable Peter Mika Researcher, Data Architect Yahoo! Research

2. - 2 - Yahoo! Research (research.yahoo.com)

3. - 3 - Yahoo! Research Barcelona Established January, 2006 Led by Ricardo Baeza-Yates Research areas –Web Mining content, structure, usage –Distributed Web retrieval –Multimedia retrieval –NLP and Semantics

4. - 4 - Yahoo! by numbers (April, 2007) There are approximately 500 million users of Yahoo! branded services, meaning we reach 50 percent – or 1 out of every 2 users – online, the largest audience on the Internet (Yahoo! Internal Data). Yahoo! is the most visited site online with nearly 4 billion visits and an average of 30 visits per user per month in the U.S. and leads all competitors in audience reach, frequency and engagement (comScore Media Metrix, US, Feb. 2007). Yahoo! accounts for the largest share of time Americans spend on the Internet with 12 percent (comScore Media Metrix, US, Feb. 2007) and approximately 8 percent of the world’s online time (comScore WorldMetrix, Feb. 2007). Yahoo! is the #1 home page with 85 million average daily visitors on Yahoo! homepages around the world, an increase of nearly 5 million visitors in a month (comScore WorldMetrix, Feb. 2007). Yahoo!’s social media properties (Flickr, delicious, Answers, 360, Video, MyBlogLog, Jumpcut and Bix) have 115 million unique visitors worldwide (comScore WorldMetrix, Feb. 2007). Yahoo! Answers is the largest collection of human knowledge on the Web with more than 90 million unique users and 250 million answers worldwide (Yahoo! Internal Data). There are more than 450 million photos in Flickr in total and 1 million photos are uploaded daily. 80 percent of the photos are public (Yahoo! Internal Data). Yahoo! Mail is the #1 Web mail provider in the world with 243 million users (comScore WorldMetrix, Feb. 2007) and nearly 80 million users in the U.S. (comScore Media Metrix, US, Feb. 2007) Interoperability between Yahoo! Messenger and Windows Live Messenger has formed the largest IM community approaching 350 million user accounts (Yahoo! Internal Data). Yahoo! Messenger is the most popular in time spent with an average of 50 minutes per user, per day (comScore WorldMetrix, Feb. 2007).  Nearly 1 in 10 Internet users is a member of a Yahoo! Groups (Yahoo! Internal Data). Yahoo! is one of only 26 companies to be on both the Fortune 500 list and the Fortune’s “Best Place to Work” List (2006).

5. - 5 - Agenda The Annotated Web SearchMonkey –Demo –Technology DataRSS format Query language Lessons learned Toward Semantic Search BOSS –Build your Own Search Service Y!OS 1.0 –Yahoo! Open Strategy

6. The Annotated Web

7. - 7 - Previously in search Horizontal search –Yahoo… –Keyword-based indexing –Minimal natural language processing –Limited experiments with ontologies (query expansion) Vertical search –e.g. shopping.com, Kelkoo –Faceted search, browsing –Fixed ontology Combinations –Google Base, Google Co-op Web-scale, but fixed ontologies Proprietary technology Can we do better with the Semantic Web? –Address the long tail of queries (88% of queries) –Use standard technology Not a new question. But the answer may be new.

8. - 8 - Which Semantic Web? Two visions –Data Web Bringing the content of databases to the Web (linkeddata.org) Rich data, heavyweight semantics Deep Web –Annotated Web Annotating the content of Web resources (documents, mm) Simple data, lightweight semantics Shallow Web This presentation is about the Annotated Web.

9. - 9 - Brief history of the Annotated Web 1995: HTML meta tags 1996: Simple HTML Ontology Extensions (SHOE) 1998: RDF/XML –RDF/XML in HTML –RDF linked from HTML 2003: Web 2.0 –Tagging –Microformats –Metadata in Wikipedia –Machine tags in Flickr 2005: eRDF 2008: RDFa

10. - 10 - HTML meta tags <LINK rel="meta" type="application/rdf+xml" title="FOAF" href= "http://www.cs.vu.nl/~pmika/foaf.rdf"> …

11. - 11 - SHOE example (Hefflin & Hendler, 1996) My name is George Cook and I live at...

12. - 12 - SHOE system

13. - 13 - SHOE Text-based query interface

14. - 14 - SHOE Graphical Query Interface

15. - 15 - Example: Creative Commons Embedding CC license in HTML (now deprecated): … … <!–- <rdf:RDF xmlns="http://creativecommons.org/ns#" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"> The Law of Averages...because eventually i&apos;ll be right... -->

16. - 16 - Example: Creative Commons Current: rel attribute (HTML4) This work is licensed under a Creative Commons Attribution 3.0 United States License. Use of the “rel” attribute for semantic annotation is the birth of the microformat…

17. - 17 - Example: microformats <a class="fn url" rel="friend colleague met" href="http://meyerweb.com/">Eric Meyer wrote a post ( Tax Relief ) about an unintentionally humorous letter he received from the Internal Revenue Service. Joe Friday +1-919-555-7878 Area Administrator, Assistant

18. - 18 - microformats microformats.org Originated by Tantek Celik and others Agreements on the way to encode certain kinds metadata in HTML –Reuse of semantic-bearing HTML elements –Based on existing standards –Community process –Persons, events, listings etc. but also syntactic metadata: licenses, tags Microformats have no shared syntax –Each microformat has a separate syntax tailored to the vocabulary Microformats are not ontologies –No formal descriptions of schema, only text –Limited reuse, extensibility of schemas –No datatypes No namespaces, unique identifiers (URIs) –no interlinking –mapping between instances is required Relationship to page context is unclear Widely used in millions of documents –User-generated as well as automatically generated

19. - 19 - Example: tags and machine tags

20. - 20 - Example: Tags and machine tags Tags –User defined keywords –Minimal agreement Is ‘rock’ on Flickr same as ‘rock’ on myspace? Is ‘rock’ by me on Flickr is the same as ‘rock’ by you on Flickr? Is ‘rock’ by me on Flickr today the same as ‘rock’ by me on myspace tomorrow? Machine tags –User defined values for user defined properties –Possibility to define the namespace (but not enforced) –Limited use

21. - 21 - RDF-based annotation #1: eRDF eRDF –Ian Davis (Talis) –Embedding RDF in HTML Straightforward mapping to RDF triples (XSLT available) HTML4 compatible –More complex than microformats Use any RDF/OWL vocabulary Reuse of semantic-bearing HTML elements is limited –More limited than RDF No blank nodes No data types No statements about subjects other than the current document –Limited usage

22. - 22 - RDF-based annotation #2: RDFa RDFa –World Wide Web Consortium (W3C) last call document –Similar intent as eRDF, but full RDF support Requires XHTML –Big question: user complexity (  data quality) Jo Smith. Web hacker at Example.org. You can contact me via email....

23. SearchMonkey

24. - 24 - Creating an ecosystem of publishers, developers and end- users –Motivating and helping publishers to implement semantic annotation –Providing tools for developers to create compelling applications –Focusing on end-user experience Rich abstracts as a first application Addressing the long tail of query and content production Standard Semantic Web technology –dataRSS = Atom + RDFa –Industry standard vocabularies http://developer.yahoo.com/searchmonkey/ SearchMonkey

25. - 25 - BeforeAfter an open platform for using structured data to build more useful and relevant search results What is SearchMonkey?

26. - 26 - image deep links name/value pairs or abstract Enhanced Result

27. - 27 -YAHOO! CONFIDENTIAL | 27 Infobar

28. - 28 - Acme.com’s database Index RDF/Microformat Markup site owners/publishers share structured data with Yahoo!. 1 consumers customize their search experience with Enhanced Results or Infobars 3 site owners & third-party developers build SearchMonkey apps. 2 DataRSS feed Web Services Page Extraction Acme.com’s Web Pages SearchMonkey

29. - 29 - Developer tool

30. - 30 - Developer tool

31. - 31 - Developer tool

32. - 32 - Developer tool

33. - 33 - Developer tool

34. - 34 - Gallery

35. - 35 - Example apps LinkedIn –hCard plus feed data Creative Commons by Ben Adida –CC in RDFa

36. - 36 - Example apps. II. Other me by Dan Brickley –Google Social Graph API wrapped using a Web Service

37. - 37 - DataRSS <feed xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.w3.org/2005/Atom../latest/xsd/datarss.xsd“> http://www.linkedin.com/datarss/ Peter Mika (pmika@yahoo-inc.com) Example data feed for social 2007-11-14T04:05:06+07:00 Peter Mika http://www.linkedin.com/ppl/webprofile?id=5054019 2007-11-14T04:05:06+07:00 John Doe male Jane Doe female Atom 1.0 XML + RDFa

38. - 38 - The data part <adjunct version="1.0" id=“com.yahoo.page.rdfa" xmlns="http://search.yahoo.com/datarss/“ updated=“2007-11-14T04:05:06+07:00”> John Doe male Jane Doe female

39. - 39 - DataRSS An Atom extension for structured data Why a new format? –A feed format is required by publishers Exclusive content (e.g. partnerships, paid inclusion) No changes necessary to the web page No standard named graph format for the Semantic Web –Needed to capture meta-metadata such as source and timestamp of information –Not really a new format An Atom extension Use any RDFa parser to get the triples out cf. Google Base feeds

40. - 40 - What happened since the launch? It’s starting to work! –Click rates improve  Publishers are willing to invest  More structured data  More applications  More users  Click rates improve Increasing excitement all around –Standardization of RDFa is bringing new energy –Good market for companies that help publishers to ‘semantify’ or support developers in extracting structured from web pages OpenCalais, Dapper, AdaptiveBlue, Intel MashMaker, Zemanta… There have been some lessons learned…

41. - 41 - Data quality Publishers/developers want the quick and dirty answer, not the long and clean one Resource or literal? – http://www.example.org – 0034691792522 Webpage or resource? –Should we allow a resource have the same URI as an existing webpage? –This is the default in eRDF/RDFa! Peter Mika Types vs. datatypes – pmika@cs.vu.nl Extensibility –rdfs:movies Complexity of the formalism = Data quality down

42. - 42 - Vocabularies Coverage is small –Books, movies, stuff people care about… Competing proposals –Versions floating around Not maintained –I cannot maintain your vocabulary for you Vocabularies for microformats –A must The role of the W3C –Ontologies as member submission…. Vocabularies not designed for the annotated Web Distributed ontology development = Mess

43. - 43 - eRDF Difficult for complex pages and dangerous in non-expert hands –Serious limitations No datatypes No subjects other than identifiers within the current page Reuse of the id attribute Peter Mika pmika@hotmail.com ….

44. - 44 - RDFa A huge improvement –E.g. no repurposing of HTML attributes Still, not everything is intuitive to the uninitiated: Peter Mika jpg … </span Peter Mika jpg </span

45. Semantic Search

46. - 46 - Semantics and IR Hard searches that cannot be solved with purely syntactical approach –Ontologies in IR shown to work in limited domains –In Web IR most attempts (e.g. query expansion) have failed What is new? The scale and breadth. –Growth in annotations, all domains (Web 2.0) –Data Web vs. Deep Web

47. - 47 - Hard searches Ambiguous searches –Paris Hilton Multimedia search –Images of Paris Hilton Imprecise or overly precise searches –Publications by Jim Hendler –Find images of strong and adventurous people (Lenat) Searches for descriptions –Search for yourself without using your name –Product search (ads!) Searches that require aggregation –Size of the Eiffer tower (Lenat) –Public opinion on Britney Spears Queries that require a deeper understanding of the query, the content and/or the world at large –Note: some of these are so hard that users don’t even try them any more

48. - 48 - Example…

49. - 49 - Application: query intent Paris Hilton is a person!

50. - 50 - Application: query intent #2 Hugo is a person!

51. - 51 - Time to experiment! Future work in IR –Ranking documents –Ranking resources/triples Ranking resources in the presence of documents –New interfaces Query interface Result presentation Metrics –Precision/recall –CTR –ROI

52. - 52 - Time to experiment! #2 Future work in Semantic Web –(Semi-)automated ways of metadata creation (NLP!) –Entity resolution –Scale –Data quality We allow providing metadata for other people’s sites! –Reasoning To the extent that it’s useful Constraints –Keyword-based search can not suffer Preference for very conservative solutions –Metadata in the context of documents Assumption is still that the user will want to see/visit the source (what to do with linked data?)

53. - 53 - Going even further… NLP, Information Extraction New types of application –Aggregators –Stateful apps –Intent-driven apps –Mobile apps –… Beyond search –Analysis, design, diagnosis etc. on top of aggregated data Personalization –Building rich user profiles –Re-ranking results based on interests Monetization –No more “buy virgins on eBay”

54. - 54 - BOSS: Build your Own Search Service Unlimited queries per day Ability to re-order results and blend-in addition content No restrictions on presentation No branding or attribution Access to multiple verticals (web search, image, news) Ability to monetize 40+ supported language and region pairs http://developer.yahoo.com/search/boss/

55. - 55 - Y!OS 1.0 Yahoo! Open Strategy –Yahoo! Social Platform Profiles, Connections, Updates, Contacts and Status –Yahoo! Query Language (YQL) Access (other) web services using a SQL-like language –Yahoo! Application Platform (YAP) Developer hosted execution of applications with access to Yahoo's Social APIs and YQL; Support for OpenSocial's JavaScript API; andOpenSocial's JavaScript API Support for server-side YML tags. Future: run applications on Y! sites OpenID, OAuth

56. - 56 - Contact Peter Mika –pmika@yahoo-inc.compmika@yahoo-inc.com –Come to Barcelona and stop by SearchMonkey –developer.yahoo.com/searchmonkey/ –mailing lists searchmonkey-developers@yahoogroups.com searchmonkey-siteowners@yahoogroups.com –forums http://suggestions.yahoo.com/searchmonkey http://suggestions.yahoo.com/datarss –Semantic Web FAQ http://devel.yahoo.com/searchmonkey/smguide/faq.html

57. - 57 - the monkey is out!