Selling Ebooks with RDF

This is the story of how O'Reilly Media began using RDF, SPARQL, and Pyramid to sell $4 million in Ebooks. Learn from our missteps, disasters, and Really Bad Ideas(TM).

Transcript

1. Delivering Ebooks with Pyramid and RDF Gavin Carothers @gcarothers Illustrations by @wholewheattoast
3. ... it begins ...
4. ... with Content Reuse!
5. • <xml/> • DocBook • XSLT • XQuery • AtomPub • Python already had:
6. ... wanted to: • Sell Chapters! • Digitize Backlist! • Make custom books! • Make textbooks! • Do Stuff!™
7. What color spine does Learning Python have?
8. What color spine does Learning Python have?
9. <xml/>
10. <tag> Soup • Tightly coupled to database • verbose • imprecise product model • slow • expensive to change
11. <xml/>
12. ONIX • Industry Standard! • Based on EDI • Ebooks added as after thought • Highly dependent on &Entities; • Unqueryable
13. ONIX • Industry Standard! • Based on EDI • Ebooks added as after thought • Highly dependent on &Entities; • Unqueryable <b203> ... title text?!
14. <xml/>
15. Design our own! • Already knew Relax NG • hundreds of hours of work • had just seen process for metadata in DocBook 5 • We ran away quickly
16. <xml/>
17. Dublin Core
18. Dublin Core Friend of a Friend
19. Dublin Core Friend of a Friend FRBR
20. Dublin Core Friend of a Friend MARC Relators FRBR
21. <xml/>
22. <xml/>
23. RDF
24. Resource Description Framework is: • Dublin Core (and other vocabularies)
25. ebook ebook ebook ebook book book book URI http://ora.com/ebook/6759 http://ora.com/ebook/8643 http://ora.com/ebook/9365
26. book POST URI book GET URI
27. Resource Description Framework is: • Dublin Core (and other vocabularies) • Document (Graph) Oriented
28. Price is what you pay. Value is what you get. —Warren Buffett
29. XQuery? doc(//om:Product[@rdf:type=‘om:Book’]/dc:creator/ @rdf:resource)/foaf:name[@xml:lang=’en’]
30. doc(//om:Product[@rdf:type=‘om:Book’]/dc:creator/ @rdf:resource)/foaf:name[@xml:lang=’en’]
31. <xml/>
32. It’s Triples!
33. subject, predicate, object . @base <http://ora.com/products/> . @prefix dc: <http://purl.org/dc/terms/> . @prefix authors: <http://ora.com/authors/> <0636920020547> dc:title “Learning SPARQL”@en; dc:creator authors:bob . @base <http://ora.com/authors/> . @prefix foaf: <http://xmlns.com/foaf/0.1/> . <bob> foaf:name “Bob DuCharme” .
34. SPARQL SELECT ?book ?name WHERE { ?book dc:creator ?author . ?author foaf:name ?name . }
39. Resource Description Framework is: • Dublin Core (and other vocabularies) • Document (Graph) Oriented • NOT <xml/> • JOINs are in the white space
43. def get_authors(product) => list product[“dc:creator”] => set
44. def customer_title(product) product[“om:customerTitle”]
45. def display_description(product) product[“dc:description”, “en”]
47. Resource Description Framework is: • Dublin Core (and other vocabularies) • Document (Graph) Oriented • NOT <xml/> • JOINs are in the white space • Simple enough to explain to designers • Fits with Python generators and dicts
48. You are going to invent at least half of it anyway.
49. Resource Description Framework is: • Dublin Core (and other vocabularies) • Document (Graph) Oriented • NOT <xml/> • JOINs are in the white space • Simple enough to explain to designers • Fits with Python generators and dicts
50. What the heck did that have to do with Pyramid?
51. Pyramid made an RDF based site POSSIBLE
52. • ... doesn’t assume you have an ORM • Or even a database at all! • ... never dictates URL structure • ... templating decoupled from framework Pyramid...
53. ... so did it work?
54. What color spine does Learning Python have?
55. What color spine does Learning Python have? X
56. $4+ million in ebook sales in first year
57. $4+ million in ebook sales in first year ✓
58. ? @gcarothers Illustrations by @wholewheattoast