Slawski New Approaches for Structured Data:Evolution of Question Answering
Download as PPTX, PDF9 likes10,863 views
Bill Slawski discusses the evolution of structured data and question-answering at Pubcon, highlighting the transition from link graphs to entity graphs, and the development of knowledge graphs. He explains the importance of annotations, rich results, and structured snippets in enhancing search engine results through schema markup. The presentation also covers the automatic discovery of entities and the use of natural language processing to improve question-answering capabilities.
Slawski New Approaches for Structured Data:Evolution of Question Answering
- 1. #pubcon Presented by: Bill Slawski (@bill_slawski) Director of SEO Research at Go Fish Digital New approaches for Structured Data: Evolution of Question-Answering
- 3. #pubcon@bill_slawski Not This Web Photo By Robert Anasch At Unsplash
- 4. #pubcon@bill_slawski Not This Web, Anymore - A Link Graph
- 5. #pubcon@bill_slawski But Now This Web - An Entity Graph • Tuple = Object/Verb/Subject • Planet of the Apes Rated G • Planet of the Apes Released in 1968 • Planet of the Apes Directed by Franklin Schaffner • Planet of the Apes has actor Roddy McDowall • Planet of the Apes has actor Charlton Heston • Planet of the Apes has actor Kim Hunter
- 7. #pubcon@bill_slawski Sergey Brin Filed a patent in 1999 on a way of crawling facts on the Web (Google’s 2nd Patent after PageRank?)
- 8. #pubcon@bill_slawski Extracting Patterns and Relations from the World Wide Web Did Sergey Want to Build Amazon?
- 9. #pubcon@bill_slawski Annotation Framework (Browseable Fact Repository=1st Knowledge Graph) Annotation Framework https://patents.google.com/patent/US20070198499A1 Patent Filed 8/2007
- 10. #pubcon@bill_slawski Annotations • Annotations are passed to requesting objects in response to queries and are used to determine search results. Annotations are also used to decide whether facts match and whether facts contain reasonable values.
- 11. #pubcon@bill_slawski MetaWeb’s FreeBase – June 2010
- 12. #pubcon@bill_slawski Knowledge Graph – May 2012
- 13. #pubcon@bill_slawski Schema.org –June 2011 • You Can Subscribe to the Public Schema Mailing List at: https://lists.w3.org/Archives/P ublic/public-schemaorg/ • To get involved in discussions and see information about monthly updates of schema.org. It is a good way to keep on top of one of the fastest growing areas of SEO
- 14. #pubcon@bill_slawski Rich Results • The Google Developer pages detail information about how to get rich snippets in search results https://developers.google.com/ search/reference/overview • Rich Results are a carrot (reward) for including Schema Markup on your pages. Google provides specific implementation details for the Schema types listed on the next page. • They try to influence you to show richer search results which can help your content stand out in SERPs, and earn more clicks.
- 15. #pubcon@bill_slawski Types of Rich Results • Article How-To Review Snippet • Breadcrumb Job Posting Sitelinks Searchbox • Book Job Training Software App • Carousel Livestream Speakable • Corporate Contact Local Business Listing Course Subscription and Paywalled Content • Critic Review Logo Video • Dataset Movie • Employer Aggregate Rating Occupation • Event Product • Fact Check QnA • FAQ Recipe
- 16. #pubcon@bill_slawski Testing Tools • The Google Structured Data Testing Tool can be found at: https://search.google.com/structured- data/testing-tool/u/0/ The Google Rich Results test can be found at: https://search.google.com/test/rich-results • Google Search Console reports on:
- 17. #pubcon@bill_slawski Table Data is Structured Data • Applying WebTables in Practice https://static.googleuserco ntent.com/media/research .google.com/en//pubs/arc hive/43806.pdf • Ten Years of WebTables • https://web.eecs.umich. edu/~michjc/papers/p2140 -cafarella.pdf
- 18. #pubcon@bill_slawski Structured Snippets • Introducing Structured Snippets, now a part of Google Web Search - https://ai.googleblog.com/2014/09/introducing- structured-snippets-now.html
- 20. #pubcon@bill_slawski Structured Data & Entities in Augmented Queries
- 21. #pubcon@bill_slawski The Evolution of Featured Snippets The First Step: Just the facts, fast Google Q&A - 2005
- 22. #pubcon Photo by Johannes Plenio on Unsplash
- 23. #pubcon@bill_slawski Facts From the Browseable Fact Repository • Extracting Facts from Documents (1) Extract facts, i.e., (subject, attribute, object) triples, from webpages to identify values of attributes, i.e., “objects” in the extracted triples. (2) Learn about patterns associated with those facts and attributes (3) Score Additional Facts from the Webpages
- 24. #pubcon@bill_slawski Answer Stores Natural Language Search Results for Intent Queries
- 25. #pubcon@bill_slawski Natural Language SERPs for Intent Queries • 1. Get Intent Questions from Authoritative Sources • 2. Get Intent Questions from Query Logs • 3. Convert Intent Questions into templates: a. What are the symptoms of XXXXXX b. What is the Treatment for XXXXXXXXXX • 4. Question and Answer Data Store is collected from authoritative sites Photo by Johann Siemens on Unsplash
- 26. #pubcon@bill_slawski Questions in Headings w/ Answers • Scoring candidate answer passages Granted 4/10/2018
- 27. #pubcon@bill_slawski Questions W/Topic Answers in Text Passages + Structured Data • Candidate answer passages Granted 1/15/2019
- 28. #pubcon@bill_slawski Interpreting Intent Behind Questions Evaluating semantic interpretations of a search query Granted July 16, 2019 Original ambiguous query: How long is Harry Potter? Semantic Interpretation: How long is the Book Harry Potter? Semantic Interpretation: How long is the movie Harry Potter? Semantic Interpretation: How tall is the character Harry Potter? Semantic Interpretation: How old is the character Harry Potter? Format = Template Question (entity) Image by Kristin Mitchell
- 29. #pubcon@bill_slawski The Evolving Knowledge Graph • Computerized systems and methods for extracting and storing information regarding entities Extract Entities, Related Entities and Entity Classes and Attributes of Entities Save Information in Tuples: Object/Verb/Subject Generate Association Scores between Entities and Attributes Based upon Factors
- 30. #pubcon@bill_slawski Association Scores • An association score may reflect a likelihood or degree of confidence that an attribute, attribute value, relationship, class hierarchy, designated context class, or other such association is valid, correct, and/or legitimate.
- 31. #pubcon@bill_slawski Association Scores in Knowledge Graph • A degree of Confidence that something is accurate or more likely to be true.
- 32. #pubcon@bill_slawski Weights Behind Association Scores • Association scores incorporate weights for each occurrence of an entity or context. • temporal weights (recent documents or occurrences) • reliability weights (more reliable sources, more heavily) • popularity weights (more popular sources, more heavily) • proximity weights (entities/contexts occurring in closer proximity to one another on a page, more heavily) Photo by Victor Freitas on Unsplash
- 33. #pubcon@bill_slawski How the Knowledge Graph Grows • Automatic discovery of new entities using graph reconciliation
- 34. #pubcon@bill_slawski Question Answering Using Knowledge GraphsNatural Language Processing With An N-Gram Machine Granted May 2, 2019 1. Use a question as a Query 2. Collect the SERPs from that Query 3. Break Pages from those SERPs into tuples 4. Build a Knowledge Graph From those tuples 5. Answer the Question Photo by Roman Kraft on Unsplash
- 36. #pubcon@bill_slawski Thank You • Bill Slawski • Author at SEObythesea.com • Director of SEO Research at Go Fish Digital • Twitter: @bill_slawski Photo by Bill Slawski, Cardiff Tidal Pools