Search and Recommendation: 2 sides of the same coin?

Loading...

Flash Player 9 (or above) is needed to view presentations.
We have detected that you do not have it on your computer. To install it, go here.

0 comments

Post a comment

    Post a comment
    Embed Video
    Edit your comment Cancel

    2 Favorites

    Search and Recommendation: 2 sides of the same coin? - Presentation Transcript

    1. TELEFÓNICA Research April 2008 Search and Recommendation: two sides of the same coin? a.k.a. Is the future of Search in Recommendation? © 2007 Telefónica Investigación y Desarrollo, S.A. Unipersonal Xavier Amatriain Researcher Where is the content?
      • 01 Introduction
      • 02 Recommender Systems
      • 03 Search Engines
      • 04 Are Search Engines going towards RS?
      • 05 Conclusions
      Index
    2. Information overload “People read around 10 MB worth of material a day, hear 400 MB a day, and see one MB of information every second” The Economist, November 2006
    3. The Age of Search has come to an end
      • ... long live the Age of Recommendation!
      • In an article published in CNN Money, entitled “The race to create a 'smart' Google”, Fortune magazine writer Jeffrey M. O'Brien, writes:
        • “The Web, they say, is leaving the era of search and entering one of discovery. What's the difference? Search is what you do when you're looking for something. Discovery is when something wonderful that you didn't know existed, or didn't know how to ask for, finds you.”
        • (Extracted from ACM Recsys 08 website)
    4. Recommender Systems in Telefonica Multimedia Entertainment E-commerce Social Networking News/Blogs/Portals Comunidades PLATFORM PRODUCTS AND SERVICES COMMERCIALIZATION Content Packaging and Design Devices Access Commercialization Customers Recommendation Systems
    5. Access through PC for exploring the whole application features Browse, Search, Rate, Comment… Get recommended series Get recommended games Meet and interact with other people Use a little application for mobile devices Get recommendations What series I watch now? Browse across TuSerie with another devices and discover new user experiences Cross-platform recommendation systems
    6. Social Networks and Recommendations Browse through user profile to check the compatibility with a specific user See the last events or the last interactions that your friends have made with the system Send messages and make this user a friend or discard him Check your compatibility with another user. View his ratings and compare them with yours. Search users
    7. Domain-specific recommendation
      • User profiling
        • Implicit through log analysis
        • Explicit asking for user feedback
      • Integrate search and recommendation in user-friendly interfaces
      Usuario Genérico User Profile Global Search Content Based Collaborative
    8. Investment in Recommendation Companies
    9. 02 Recommender Systems
    10. The “Recommender problem”
      • Estimate a utility function that is able to automatically predict how much a user will like an item that is unknown for her. Based on:
        • Past behavior
        • Relations to other users
        • Item similarity
        • Context
        • ...
      Offline Online
    11. Approaches to Recommendation
      • Collaborative Filtering
        • Recommend items based only on the users past behavior
          • Similarity between users or items computed only from this
        • User-based
          • Find similar users to me and recommend what those users liked
        • Item-based
          • Find similar items to those that I have previously liked
      • Content-based
        • Recommend based on features inherent to the items
    12. Recommendation as a Datamining problem
      • The core of the Recommendation Engine has been assimilated to a general datamining problem:
      • However RS have attracted input from a large community
        • IR, e-Commerce, HCI, Psychology...
    13. What works
      • It depends on the domain: Domain-specific modeling
      • However, in the general case it has been demonstrated that (currently) the best isolated approach is CF.
        • Item-based in general more efficient and better but mixing CF approaches can improve result
        • Other approaches can be hybridized to improve results in specific cases (cold-start problem...)
      • What matters:
        • Data preprocessing: outlier removal, denoising, removal of global effects (e.g. individual user's average)
        • “Smart” dimensionality reduction using MF such as SVD
        • Combining classifiers
    14. Data mining + all those other things
      • User Interface
      • System requirements (efficiency, scalability, privacy....)
      • and ....
    15. Serindipity
      • Unsought finding
      • Don't recommend items the user already knows or would have found anyway .
      • Expand the user's taste into neighboring areas by improving the obvious
      • Collaborative filtering can offer controllable serendipity (e.g. controlling how many neighbors to use in the recommendation)
    16. 02 Search
    17. Search Engines
      • General process diagram of a search engine
      Offline Online
    18. Search vs. recommendation
      • Is search a content-based “recommendation”?
        • In the indexing and retrieval processes we are trying to “cluster” similar documents based exlusively on content (no user information)
      • or a poor-man's approach to CF?
        • Most ranking algorithms can be seen as a simplified collaborative filtering where we are recommended the opinion of the average user's (what most people link) or the authorities (e.g. Page Rank).
        • To some extent we can say that web “structure” reflects past users' behavior
    19. 03 But, is Search going towards Recommendation?
    20. Personalized Search
      • Last year's presentation in this same Workshop is a good starting point (Paul-Alexandru Chirita, Current Approaches to Personalize Web Search)
      • Overall trend -> Use personalized user profile in order to improve returned page ranking
    21. Recent advances in Personalized Search
      • Interesting approaches:
        • Automatic Identification of User Interest for Personalized Search (Qiu et al WWW06)
          • Improve topic-sensitive page rank by inferring topic preference vector for the user.
          • Very similar to content-based recommendation
        • CubeSVD: A Novel Approach to Personalized Web search (Sun et al WWW05)
          • LSI using HOSVD to find a score for webpages based on q,u pairs.
          • Very similar to CF
    22. A hybrid Search-Recommender? == ? Recommendation Search
    23. 04 Conclusions
      • The ever-growing amount of content makes searching difficult (time-consuming and unsatisfactory)
        • Too much to search for, too many results
        • Frustration from it not being adaptive
      • Search is starting to take the user into account
      • Is search something users want to do or just something they can do with the tools we offer?
      • Are search and recommendation two sides of the same coin?
        • Is search about retrieval and recommendation about ranking?
        • Should they complement each other or become the same thing?
    24. [email_address] http://www.tid.es We're hiring!
    SlideShare Zeitgeist 2009

    + Xavier  AmatriainXavier Amatriain Nominate

    custom

    205 views, 2 favs, 0 embeds more stats

    Invited talk at the Workshop on the Future of Searc more

    More info about this document

    © All Rights Reserved

    Go to text version

    • Total Views 205
      • 205 on SlideShare
      • 0 from embeds
    • Comments 0
    • Favorites 2
    • Downloads 21
    Most viewed embeds

    more

    All embeds

    less

    Flagged as inappropriate Flag as inappropriate
    Flag as inappropriate

    Select your reason for flagging this presentation as inappropriate. If needed, use the feedback form to let us know more details.

    Cancel
    File a copyright complaint
    Having problems? Go to our helpdesk?

    Categories