Mais conteúdo relacionado




Linked Data: The Real Web 2.0 (from 2008)

  1. Linked Data: The Real Web 2.0 Uche Ogbuji <> 17 June 2008, NYC
  2. Warming up...
  3. Hi. I’m Uche Ogbuji • Partner at Zepheira • • Web architecture & Semantic technology solutions Ask me about: START, 3D, Remix • Longtime leader in XML, data integration and semantic technology • Lead on key open specs & FLOSS, e.g. • • • Jacquard Web development methodology Versa RDF query language 4Suite & Amara XML/RDF tools Nigerian-American. Father. Husband. Poet. Entrepreneur. Engineer. Tech pundit. Footballer (“soccer”). Snowboarder. Skateboarder. Hip-Hopper.
  4. The goal: enough Linked/Linking (Open) Data basics for the curious But enough deep discussion for the initiated
  5. How we got here
  6. Who wants to repeat Web history? OK then let’s follow Ken North’s lead and study it...
  7. A brief history... • The Web, mid-1990s • Big daisy chain of home pages. Not much need for context in hyperlinks. • Sloppy was OK. Remember 404s? • Then it started to get big. Portal wars, search engine wars, browser wars... • Sloppy was starting to really hurt...
  8. First wave solutions • XML for the document weenies. RDF for die echten menschen • Hey, if the W3C builds it, they will come • Semantic Web FTW! • Time to sort out the chaos of the Web. Context by formal declaration. • Anyone here tried explaining OWL to Joe Webmaster?
  9. Back to architecture • Enter SOAP Web services: suddenly we had more pressing concerns • Uh oh! There’s a cuckoo’s egg in the Web • Phew! Got that anti-cuckoo detection kit ready just in time • We fought back with REST • Lesson learned: if we didn’t keep to the original, simple principles
  10. What’s that lesson again? • We think we know what makes the Web work so well • We want others to understand, so we want to have suggestions and tools for them • We’d better keep to the simplest suggestions and tools that could possibly serve our cause
  11. Simplest that could possibly... • So Semantic Web, right? • You kidding? Ever tried to explain SemWeb to Joe Webmaster? • You kidding? A Ph.D. dissertation? Get Joe Webmaster the Cliff Notes • So... REST, right? • OK. OK. I give up...
  12. Spoiler alert!
  13. Tim Berners-Lee breaks it down for the oh-six
  14. Back to history: Web 2.0 • Bored one day, The Weblog echo chamber dusted off some old Web tricks • Enough smart developers came up with enough simple tools and suggestions to create a craze • So is this a catastrophic distraction from what the Web really needs?
  15. What does Web 2.0 really boil down to? “Think globally, act locally”
  16. Say what? • You make some modest enhancements to your site (the local). That’s all you need to worry about. • You then roll in the global power of Web-enabled services: Mashups, feeds, user-generated content... • ...Mumble...SEO...mumble...long tail...mumble... • ...Profit • What Webmastering Joe could resist that call?
  17. But hey! But wait! But SWEET! “Think global, act local” is all we needed from Joe anyway, right? You know, to make a better Web????!!!
  18. SCORE!!!!
  19. Web2: Web feeds • RSS, Atom, JSON... • For a bunch of non SemWebby guys Atom 1.0 community did a decent job • Finally gives us a reasonably pervasive format for reading, with some true context retained • A small step towards a machine-readable Web • Small steps matter! • Read the Web
  20. Web2: User-generated content • Hey, isn’t this the write part of the read-write Web we’ve been craving? • Thank Wikipedia just for the psychological fillip for the non-techie: you too can write • Sure forums technically offered that for ages, but Wikipedia carries an clear, broadcast message. It’s prime time. • Write the Web • Oh BTW Remember that Web of Trust thingie SemWebbers were chattering up years ago? • Psssssst! We kinda really need that now. Thanks.
  21. Web2: Mashups • “So, like, all I have to do is focus on the sort of information I specialize in, and mark it up in clear context” • “Yep. And other smart folks just combine it with other data in ways you never imagined” • “OK, that’s worth paying some attention to Web data design”
  22. SCORE!!!!
  23. Web2: Mashups • Joe’s sophistication might end with mashing, but a good data architect can make meshes of mashes • Remix and Re-purpose the Web
  24. Sure, sure, but all that is why we’re Web 3.0 We’re too cool for small steps!
  25. Yeah, because version number wars did so much good for... Office suites DBMS RSS
  26. We’re not different from Web 2.0 We’re just Web 2.0, properly • Vendor independent • Scalable • Extensible • Multi-device friendly. • It’s not just about the PC/Mac bowser, you know
  27. Marketing command: What would Web 2 do? Heresy maybe, but marketing matters, even when you’re just trying to do the right thing
  28. Give them something they can figure out in 10 minutes
  29. Linked Data
  30. Linked Data means just remembering 4 things 1. Use URIs as names for things 2. Use HTTP URIs so that people can look up those names. 3. When someone looks up a URI, provide useful information. 4. Include links to other URIs. so that they can discover more things.
  31. From kernels of wisdom grow... • Joe Webmaster can understand this, and probably remember this • #3 and #4 are openings for clever types to push for more •It’s all about the link, ya’ll
  32. <a href... • The granddaddy link • The most pervasive link • Limited for preserving context • @rel, @type and such do help • Just plain Limited.
  33. XLink? Hlink? • OK <a href is limited, but at least you can grasp it in minutes • We like our snoozefest flame wars, eh? • How many URI hashes/slashes can dance on the head of a pin? • What is the link to the divine? Sophisticated Gnosis? Simpleton Chrestos? The 3-personed markup? • Needless to say, not too many people are using XLink, HLink and friends
  34. RDF • RDF is in effect a linking technology • <a xml:base=”Subject” href=”Object” rel=”predicate”>... • Generally separate from described resources (not unlike XLink link-bases) • Also a fairly heavy conceptual burden • An unfortunate syntax is more prevalent than the useful model
  35. That will bring us back to...
  36. Yes: <a href... • Worse is better, folks • Just a few careful conventions and we can squeeze a bit more out of it • • @rel conventions @type coordinated with core HTTP (media types, content negotiation, etc.) • Let’s first see what we can build on that, then worry about the gaps
  37. Linking Open Data
  38. Clever types building on Linked Data • Emphasize a particular set of conventions and tools for refining Linked Data • HTTP usage such as content negotiation or 303s for linking to abstract resources • Syntax layer usage such as RDFa and GRDDL • Emphasize a particular set of sites as seeds for growing Linked Data • DBPedia is the star
  39. RIchard’s LOD bubble diagram
  40. What would Web 2 do?
  41. Kingsley’s four techniques 1. Handle Content Negotiation requests 2. <link rel=-"alternate"..../> (when HTML is requested) 3. GRDDL profile in <head/> plus <link rel="transformation".../> (when (X)HTML or XML is requested) 4. eRDF or RDFa (when HTML or (X)HTML is requested) Note: Taken from a community posting. See Kingsley’s presentation for his own refinement of this...
  42. This is exciting. We’re hitting our stride folks.
  43. Just one thing: Don’t forget the nuanced message for Joe CIOverlord
  44. Linking Enterprise Data
  45. What is LED? • A community project to develop tools and architectural strategies to bring Sensible SOA to the Enterprise through data-driven applications • You’ve heard of Linking Open Data? LED comes in when the data can’t always be so open • You’ve heard of Enterprise 2.0 and Enterprise Mashups? LED comes in when you need such applications to outgrow the toy box • Same relationship as with Linked Data to Web 2.0
  46. Business context • LED involves using tools to express business context as rich, reusable metadata (often RDF) • LED involves using well-known data syndications tools to orchestrate data enriched with this context • LED involves using formally expressed policy to control data flow and capture responsibility
  47. Bridge to legacy • Rather than the ERP-type play to replace legacy apps with a centralized super-model, LED focuses on wrapping and exposing data in those apps • The exposed and contextualized data from source applications is integrated using basic Web technology
  48. LED step by step • Capture business context during analysis • Capture business rules and policy during analysis • Express these in simple, sharable, reusable formats made available to people as well as code (SKOS, Attempto-Controlled English, etc.) • Map data models of existing applications to the expressed context (declarative data transform)
  49. More LED step by step • Architect Web integration layer around documented context • Use mappings to drive Web-feeds from legacy apps to the integration layer (JSON, Atom, RDF, etc) • Use lightweight so-called Web 2.0 kit to simplify browser access
  50. Questions?