1 / 21

Personalize your site by identifying your user s ' alter-egos.

Personalize your site by identifying your user s ' alter-egos. What do we do?. Provide widget-based recommendation User centric Scalable Easy to interface Outcome of an ERC Starting Grant Project (2008-2013). How do we do that ?. For each user

loki
Download Presentation

Personalize your site by identifying your user s ' alter-egos.

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Personalize your site by identifying your users' alter-egos.

  2. What do we do? • Providewidget-basedrecommendation • User centric • Scalable • Easy to interface • Outcome of an ERC Starting Grant Project (2008-2013)

  3. How do we do that? • For each user • Maintain a profile (visited pages or items bought or clicked) • Identify alter-egos on the site (or across sites) • Leverage the alter-egos profiles to filter and recommend content

  4. Mediego architecture Mediego servers Client Web site Profile table KNN table I1 I2 … Im … widget Display Recommendation for

  5. Mediego computations Mediego servers Client Web site Profile table KNN table I1 I2 … Im … widget Display Recommendation for

  6. Products • Recommendationwidgets for Web sites • Media content providers • E-commerce • Otherproducts • News recommender (Dashboard AllYours) • P2P news recommender

  7. Targets • Recommendationwidgets for Web sites • Media editors: recommendationswithoutneed for subscriptions (automaticfiltering) • E-commerce • Recommendation of productswithin a site (example: chronodrive, La redoute) • Recommendatio leveragingactivities on other sites

  8. CSATT février 2013 - AllYours Recommendation widget

  9. Targets • Recommendationwidgets for Web sites • Media editors: recommendationswithoutneed for subscriptions (automaticfiltering) • E-commerce • Recommendation of productswithin a site (example: chronodrive, La redoute) • Recommendatio leveragingactivities on other sites

  10. Where are we? • Software • operational and simple gateway • CentralizedMediego • CentralizedAllYours/P2P Allyours • Patents • One filled, 4 in the process • Fundings • ERC PoC (2013), EIT ICT (2013, probably 2014)

  11. Nextbigstep: Beta tests • Media content editors • Wamiz • Femina.ch • Jeux.video.fr • Contact le Monde • E-commerce • Carrefour • Channel • Cdiscount Live Beta tests TrentoRise • P2P • Dashboard • Students

  12. Competitors • Outbrain: Present on many media sites • Taboola • Criteo: • Amazon/predigo • Target to sell

  13. The team • PI: Anne-Marie Kermarrec • Chef de projet: Sébastien Campion (20%) • Engineers • Antoine Boutet • Jacques Falcou • Heverson Ribeiro (P2P) • Jean-FrancoisVerdonck • Others • Davide Frey, Researcher INRIA • Rachid Guerraoui, Prof. EPFL • Arnaud Jégou, PhDstudent INRIA • Marc Thouvenin (neovira) • Laurent Kott (IT Translation) • Patric Gélin/Marie-Christine Lancien (INRIA)

  14. Business models • Media • Increasetheirnumber of viewed pages (adds) • Increasetheirnumber of subscribers • E-commerce • - Increase the number of productsdiscovered/bought

  15. Business models • Media • Increasetheirnumber of viewed pages (adds) • Increasetheirnumber of subscribers • E-commerce • - Increase the number of productsdiscovered/bought

  16. On the technical side

  17. Architecture KNN (AlterEgos) User input Recommendation/ dissemination User input

  18. CSATT février 2013 - AllYours News recommender

  19. CSATT février 2013 - AllYours WhatsUp: a centralized news recommender • Architecture centralized • Centralized d ata base of users profile and iitems • Centralized Top-k algorithm to build the social network (based on sampling) and centralizedrecommendationengine • Challenges • Operational prototype • Gatherusersactivityautomatically • Metrics of success • - Gathering of a community of (satisfied) users Top-K neighborhood User widget Open API Recommendation/ dissemination

  20. CSATT février 2013 - AllYours A centralizedtool for our beta testers • Architecture centralized • Centralized data base of users profile and iitems • Centralized Top-k algorithm to build the social network (based on sampling) and centralizedrecommendationengine • Challenges • Operational prototype • Gatherusersactivityfromvarious Web sites • Metrics of success • Feedback fromour beta testers Top-K neighborhood User widget Open API Recommendation/ dissemination

  21. CSATT février 2013 - AllYours Exemple on Jeux videos.com

More Related