Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

HP's Everything as a Service Vision and HP Gloe

1,275 views

Published on

Presentation of the crowdsourced geotagging service HP gloe from a vision and engineering perspective

Published in: Technology
  • Be the first to comment

  • Be the first to like this

HP's Everything as a Service Vision and HP Gloe

  1. 1. Thomas Sandholm—Research Scientist, HP Labs, Palo Alto<br />December 17, 2009<br />HP’s Everything as a Service Vision and HP Gloe<br />
  2. 2. Everything as a Service<br />“A world of information, opportunities and experiences — from computing power to business processes to personal interactions —<br />delivered wherever, however and whenever you need it” –Shane Robison, Hewlett-Packard CTO<br />
  3. 3. Shopping in Russia<br />Social networking in Germany<br />Uploading video in the US<br />Researching in China<br />Chatting in India<br />Blogging in Brazil<br />Then: Efficiency and performance were driving innovation<br />Now: Explosion in new modes of communication and collaboration<br />Distinction between content consumers and producers blurred<br />Real-time co-active content production <br />
  4. 4. Challenges<br />Opportunities<br /><ul><li>Information explosion
  5. 5. Filter info fast enough
  6. 6. YouTube 20h/min video
  7. 7. Globalization
  8. 8. Emerging markets
  9. 9. Heterogeneity
  10. 10. Environmental sustainability
  11. 11. Reduce carbon footprint
  12. 12. IT footprint vs non-IT process footprint</li></ul>Challenges and Opportunities<br /><ul><li>Harness economics of attention
  13. 13. Money follows eye-balls
  14. 14. Max attention – Max ad revenue
  15. 15. Crowdsourcing
  16. 16. Production and vetting by crowd
  17. 17. Tap collective intelligence
  18. 18. Consumerization of IT
  19. 19. Non IT-savvy users
  20. 20. Facebook elderly demographic growth</li></li></ul><li>Intuitive technology that anticipates your needs<br /><ul><li>Information liberated from devices
  21. 21. Software-as-a-Service, e.g. Salesforce.com, Google Docs
  22. 22. Collaborative filtering, recommendations based on other users’ behavior
  23. 23. Services aware of context and location
  24. 24. Sensors determine behavior
  25. 25. Personal usage history for customization
  26. 26. Search is done for you not by you
  27. 27. Recommendations and feeds
  28. 28. Traditional key-word search tricky on mobile devices</li></li></ul><li>HP GLOE<br />
  29. 29. HP Gloe<br />Social, geo-aware Web browsing<br />http://www.hpgloe.com<br /><ul><li>Find and recommend local Web content
  30. 30. Crowdsourced mappings of Web pages to locations
  31. 31. Folksonomy tagging
  32. 32. Social filtering
  33. 33. Context = Location, Friends, Channel (tag filter) </li></li></ul><li>
  34. 34. Gloe ArchitectureRESTful JSONP and LAMP Open Source Stack<br />Client<br /><ul><li>HTTP GET input
  35. 35. JSON or HTML output
  36. 36. JavaScript (Web), Android (Java), Blackberry (Java ME), Python </li></ul>Server<br /><ul><li>Apache HTTP/S
  37. 37. Django Python</li></ul>Database<br /><ul><li>MySQL
  38. 38. GeohashSharding</li></ul>Security<br /><ul><li>HMAC/SHA1
  39. 39. HTTPS
  40. 40. SocialAuth/Facebook Connect</li></li></ul><li>Gloe Service DesignKeep It Small and Simple<br />maintenance<br />Use off-the-shelf tools/protocols/frameworks<br />Avoid custom serialization and stub generation<br />Minimal meta-data<br />Minimize on-the-wire data<br />Local lookups fast and scalable (max hits, radius)<br />De-normalized database tables<br />Don’t reinvent a social network<br />Feeds triggered by location<br />Make it easy to add contributions (single click when surfing)<br />performance<br />experience<br />
  41. 41. Gloe Protocolhttp://www.scribd.com/doc/23673122/HP-Gloe-API<br />Get all recommendations for geolocation (37.234,-122,234)<br />http://www.hpgloe.com/json/getrec/?lat=37.234&lon=-122.234<br />Recommend http://www.hp.com in geolocation (37.234,-122,234) on channel hp<br />http://www.hpgloe.com/json/rec/?lat=37.234&lon=-122.234&url=http://www.hp.com&tags=hp<br />
  42. 42. Gloe Protocol JSON Example<br />[[3.0, &quot;http://en.wikipedia.org/wiki/Middleton_Tract%2C_California&quot;, &quot;&quot;, null, null, null, 37.266834259033203, -122.208666483561, 0.0, 2.66548816497217], [3.0, &quot;http://en.wikipedia.org/wiki/National_Register_of_Historic_Places_listings_in_Santa_Clara_County%2C_California&quot;, &quot;&quot;, null, null, null, 37.264633178710902, -122.074666341146, 0.0, 9.1596029435647299], [2.0, &quot;http://en.wikipedia.org/wiki/Boulder_Creek%2C_California&quot;, &quot;&quot;, null, null, null, 37.128200531005902, -122.124500274658, 0.0, 9.4695493134283808], [2.0, &quot;http://en.wikipedia.org/wiki/KFJC&quot;, &quot;&quot;, null, null, null, 37.320598602294901, -122.140998840332, 0.0, 7.86499127030193], [2.0, &quot;http://en.wikipedia.org/wiki/Los_Trancos_Woods%2C_California&quot;, &quot;&quot;, null, null, null, 37.349399566650398, -122.198497772217, 0.0, 8.2030095137784897], [2.0, &quot;http://en.wikipedia.org/wiki/National_Register_of_Historic_Places_listings_in_San_Mateo_County%2C_California&quot;, &quot;&quot;, null, null, null, 37.126800537109403, -122.308498382568, 0.0, 8.4687380823758893], [2.0, &quot;http://en.wikipedia.org/wiki/Portola_Valley%2C_California&quot;, &quot;&quot;, null, null, null, 37.375, -122.21900177002, 0.0, 9.7701614508682404], [2.0, &quot;http://www.panoramio.com/photo/4489152&quot;, &quot;Silicon%20Valley%20View%20From%20Mountain%20Winery&quot;, null, null, null, 37.258098602294901, -122.06300354003901, 0.0, 9.5446305308779191], [1.0, &quot;http://en.wikipedia.org/wiki/A%C3%B1o_Nuevo_State_Reserve&quot;, &quot;&quot;, null, null, null, 37.133098602294901, -122.333000183105, 0.0, 8.8426858714568901], [1.0, &quot;http://en.wikipedia.org/wiki/Alpine_Creek&quot;, &quot;&quot;, null, null, null, 37.3101997375488, -122.27799987793, 0.0, 5.7900750211751397]]<br />
  43. 43. Summary<br />HP’s Everything as a Service vision: information delivery anywhere, anytime<br />Crowd content production implication – information overflow<br />HP Gloe solves information overflow with social, location, and folksonomy-tagging based filtering<br />HP Gloe Architecture REST/JSON/LAMP<br />
  44. 44. Q&A<br />

×