Ny Freebase Workshop 10 Dec 2009

1,119 views

Published on

Intro slides for NY Freebase Workshop on Dec 10, 2009

0 Comments
2 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
1,119
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
36
Comments
0
Likes
2
Embeds 0
No embeds

No notes for slide

Ny Freebase Workshop 10 Dec 2009

  1. 1. Freebase<br />New York Workshop<br />10 Dec 2009<br />
  2. 2. Presenters<br />Robert Cook<br />Jamie Taylor<br />Will Moffat<br />
  3. 3. Today’s Workshop<br /> 9:30 – Intro<br /> 10:30 – Prepackaged Freebase solutions<br /> 12:30 – Lunch<br /> 1:15 – Connecting your data to Freebase<br /> 2:30 – Freebase in the data service ecosystem<br /> 3:30 – Wrap up, “office hours”<br />
  4. 4. Agenda<br />Intro to Freebase<br />Freebase as an identity directory<br />The Freebase platform<br />
  5. 5. Metaweb<br />Technology company based in San Francisco<br />~60 person team of engineers and business people<br />Venture funded, with long-term outlook<br />Focused on Freebase.com platform<br />
  6. 6. Freebase is a database of entities<br />One entity per thing in the world<br />Stable, long-lived identifiers<br />Inclusive policy<br />/en/sienna_miller<br />Practical data<br />Focus on available data<br />People, places, products, etc.<br />/en/sony_dsc_s750<br />Data to build apps<br />Names, images, descriptions<br />Dates, measurements and relationships<br />/en/frost_nixon_2008<br />
  7. 7. Actresses (37,079)<br />
  8. 8. Football Players (16,568)<br />
  9. 9. Cheeses (488)<br />
  10. 10. Musical Instruments (1,034)<br />
  11. 11. Airports (11,556)<br />
  12. 12. TV Programs (33,630)<br />arrested_develop<br />
  13. 13. Related entities are connected, forming a graph<br />Current stats:<br /><ul><li>~10M entities
  14. 14. ~1,800 “types”
  15. 15. Celebrity
  16. 16. Movie
  17. 17. TV show
  18. 18. Book
  19. 19. Company
  20. 20. Location
  21. 21. Sports team
  22. 22. Product
  23. 23. Etc.
  24. 24. ~275M facts
  25. 25. Continuous data input, cleanup, and syncing</li></li></ul><li>Each entity contains rich, structured metadata<br />
  26. 26. Entities are language independent<br />
  27. 27. As a writeable graph, Freebase gets better over time<br /><ul><li>Add (or remove) entities
  28. 28. Add (or remove) metadata (facts, keys, translations, etc.)
  29. 29. Extend and improve the schemas</li></li></ul><li>Bulk data into Freebase<br />15 person group dedicated to algorithmic data import, processing, and tools development<br />Reconciliation, reconciliation, reconciliation<br /> Critical part of everything we do <br /> Automate wherever possible<br />Crowdsource for tasks requiring human judgment (semi-automated)<br />Pipelined, ongoing syncing with large external sources(Wikipedia, partners, etc.)<br />
  30. 30. Reconciliation<br />Guaranteeing one entity <br />per thing in the world<br />
  31. 31. Reconciliation<br />
  32. 32. Reconciliation<br />
  33. 33. Reconciliation<br />
  34. 34. “US Politicians who have taken more than $30K from foreign companies”<br />
  35. 35. Freebase is open<br />
  36. 36. Open platform means more data<br />Creative Commons Attribution(CC-BY) licensing<br />Robust set of APIs<br />HTTP/REST<br />SLAs for higher volume users (typically &gt;100K API calls per day)<br />Hosted developer platform for building tools and apps on top of the data<br />Apps<br />
  37. 37. External site data and/or keys<br />Beer (3,100)<br />The Oxford Bottled Beers Database<br />TV episode (715,032)<br />The TVDB, TV Rage, etc.<br />
  38. 38. A global community is actively improving it<br />Creating new data sets<br />Curating existing data<br />sprocketonline<br />Jet Engines<br />Hummingbirds<br />spatialed<br />tfmorris<br />Maritime museums<br />
  39. 39. The community is defining new schemas<br />Top-level domains<br />
  40. 40. Agenda<br />Intro to Freebase<br />Freebase as an identity directory<br />The Freebase platform<br />
  41. 41. Everybody is creating entities<br />Topic pages<br />User profiles<br />Relevant apps<br />Artist pages<br />Other fans<br />
  42. 42. Millions of users are helping them<br />@robcook(Person)<br />#sxsw09 (Event)<br />(Movies, Celebrities, Companies, Products, etc.)<br />
  43. 43. Freebase is connecting these entities together<br />Will Smith(Actor)<br />/index.html?curid=154698<br />/people/s/will_smith<br />/name/nm0000226<br />/RoleDisplay/86971<br />/BandsAndArtists/S/Smith,_Will<br />/artist/Will+Smith<br />willsmith.com<br />/WillSmith<br />/artist/Will+Smith<br />/music/Will+Smith<br />/Will-Smith/e/B000APUOJC<br />
  44. 44. An entity directory can power new applications<br />
  45. 45. Example: <br />Each film review is tagged with the corresponding movies in Freebase<br />TheIncredibles(film)<br />Alfie(film)<br />When the pages loads, it grabs data from Freebase (images, film info and links) to enhance the article<br />Freebase also returns links to related WSJ film reviews the user might enjoy (based on genre, director, actors, release year, etc.)<br />A Freebase search box allows the user to quickly find any film review in the WSJ archives<br />
  46. 46. Agenda<br />Intro to Freebase<br />Freebase as an identity directory<br />The Freebase platform<br />
  47. 47. Freebase architecture<br />
  48. 48. Query editor<br />
  49. 49. Querying Freebase<br />“Russian cosmonauts”<br />[{<br /> &quot;type&quot;: &quot;/spaceflight/astronaut&quot;,<br /> &quot;name&quot;: null,<br /> &quot;/people/person/nationality&quot;: ”russia&quot;<br />}]<br />
  50. 50. Querying Freebase<br />“Tropical storms in the 90s”<br />{<br /> &quot;type&quot;: &quot;/meteorology/tropical_cyclone&quot;,<br /> &quot;name&quot;: null,<br /> &quot;formed&gt;=&quot;: &quot;1990&quot;,<br /> &quot;a:formed&lt;&quot;: &quot;2000” <br />}<br />
  51. 51. Querying Freebase<br />“French actresses born pre-WWII”<br />{<br /> &quot;type&quot;: &quot;/film/actor&quot;,<br /> &quot;name&quot;: null,<br /> &quot;/people/person/gender&quot;: &quot;female&quot;,<br /> &quot;/people/person/date_of_birth&lt;=&quot;: &quot;1939&quot;,<br /> &quot;/people/person/nationality&quot;: &quot;France&quot;,<br /> &quot;sort&quot;: &quot;/people/person/date_of_birth&quot;<br />}<br />
  52. 52. ACRE<br />Server side Javascript + webpage templating<br />WSJ (and other) applications developed<br />Advanced APIs<br />Code sharing – programmer ecosystem<br />
  53. 53. ACRE IDE<br />
  54. 54. Other platform services<br />Freebase suggest<br />Lucene-based topic search interface<br />Blob store (text, image thumbnailing)<br />Reconciliation service<br />Extended MQL<br />
  55. 55. www.freebase.comblog.freebase.com<br />twitter.com/fbaserobert@metaweb.com<br />

×