hummingbird & the entity revolution

33
Hummingbird & The Entity Revolution – Knowledge and Search, Two Swords Sharpening One Another Bill Slawski SMX East 2014 (#smx #21A) October 1, 2014 (9:00am-10:15am)

Upload: bill-slawski

Post on 26-Jun-2015

2.126 views

Category:

Documents


1 download

TRANSCRIPT

  • 1. Bill SlawskiSMX East 2014 (#smx #21A)October 1, 2014 (9:00am-10:15am)

2. When Sergey Gave Larry a TourWe both found each other obnoxious, Brin counterswhen I tell him of Page's response. "But we say it alittle bit jokingly. Obviously we spent a lot of timetalking to each other, so there was something there.We had a kind of bantering thing going." Page andBrin may have clashed, but they were clearlydrawn together - two swords sharpening oneanother.The Birth of Google by John Battelle #smx #21A @bill_slawski 3. #smx #21A @bill_slawski 4. Larry Invents PageRankImproved Text Searching in Hypertext Systems (pdf) #smx #21A @bill_slawski 5. #smx #21A @bill_slawski 6. Sergey Invents DIPREExtracting Patterns and Relations from the World Wide Web (pdf)#smx #21A @bill_slawski 7. #smx #21A @bill_slawski 8. Patterns!#smx #21A @bill_slawski 9. Andrew Hogues TeamAndrew Hogues Resume #smx #21A @bill_slawski 10. Patterns!#smx #21A @bill_slawski 11. Gathering & Annotating KnowledgeBrowseable fact repository #smx #21A @bill_slawski 12. Search becomes Knowledge#smx #21A @bill_slawski 13. Googles Knowledge GraphThe Knowledge Graph #smx #21A @bill_slawski 14. #smx #21A @bill_slawski 15. Alexandria torpedo factory CC BY-SA 3.0 #smx #21A @bill_slawski 16. Google Starts a ConversationFAQ: All About The New Google HummingbirdAlgorithm#smx #21A @bill_slawski 17. Entities become Search EntitiesSearch entity transition matrix and applications of the transition matrixRelationships between Search Entities#smx #21A @bill_slawski 18. Which [lincoln]?#smx #21A @bill_slawski 19. Tracking Knowledge InformationEach record (herein referred to as a tuple: ) comprises a query submitted byusers, a document reference indicating the documentselected by users in response to the query, and anaggregation of click data for all users or a subset of allusers that selected the document reference in responseto the query.Propagating query classificationsUsing Query User Data to Classify Queries#smx #21A @bill_slawski 20. Google is Viewing Entities as Search Entities (lincoln as aperson is a query) likely using an RDF (Resource Description Framework)schema for tracking information to calculateprobabilities (based on user behavior) of things likewhat classification is meant by a query. Searching for Patterns in Queries and on Pages toanswer questions Looking for Schema markup and schema-relatedfacts and attributes information to create andunderstand context.#smx #21A @bill_slawski 21. Patterns!#smx #21A @bill_slawski 22. Query Revision Based on Contextand Substitute RulesSynonym identification based on co-occurring terms #smx #21A @bill_slawski 23. For example, the user may enter the search query "What isthe best place to find and eat Chicago deep dish stylepizza?"In determining whether the term "restaurant" is a synonymfor the query term "place", a synonym engine may evaluatethe query term in the context of adjacent terms, such as"best" or "to," as well as non-adjacent terms, such as"Chicago" and "pizza."Such an evaluation may result in the decision that, in thecontext of the non-adjacent term "pizza," the term"restaurant" is a synonym of the query term "place."#smx #21A @bill_slawski 24. Knowledge Base SearchesIdentifying entities using search results #smx #21A @bill_slawski 25. The Future of the KnowledgeGraph?Knowledge Vault: A Web-Scale Approach to Probabilistic Knowledge Fusion#smx #21A @bill_slawski 26. Incompleteness of KnowledgeGraph#smx #21A @bill_slawski 27. Introducing the Knowledge Vault?Constructing and Mining Web Scale Knowledge #smx #21A @bill_slawski 28. Recovering Semantics of Tables on the Web (pdf)#smx #21A @bill_slawski 29. Open Language InformationExtractionOpen Language Learning for Information Extraction (PDF)#smx #21A @bill_slawski 30.