1 / 51

Correlation-Based Content Adaptation For Mobile Web Browsing

Correlation-Based Content Adaptation For Mobile Web Browsing. Iqbal Mohomed, Adin Scannell, Nilton Bila, Jin Zhang, Eyal de Lara Department of Computer Science University of Toronto. Middleware 2007. Need for Adaptation. Downloaded Data  600KB. Content must be customized!.

Télécharger la présentation

Correlation-Based Content Adaptation For Mobile Web Browsing

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. Correlation-Based Content Adaptation For Mobile Web Browsing Iqbal Mohomed, Adin Scannell, Nilton Bila, Jin Zhang, Eyal de Lara Department of Computer Science University of Toronto Middleware 2007

  2. Need for Adaptation Downloaded Data  600KB Content must be customized!

  3. Automatic Adaptation Mobile Device Adaptation Proxy Unmodified Content Server We Have The Mechanism But … The Hard Problem is Policies!

  4. Achieving Fine Grain Adaptation • Usage and context both affect the customization that is needed • Machines have a hard time distinguishing between good and bad adaptations • People are better

  5. Achieving Fine Grain Adaptation • Usage and context both affect the customization that is needed • Machines have a hard time distinguishing between good and bad adaptations • People are better Rely on a few users to adapt content for everyone!

  6. Usage-awaRe Interactive Content Adaptation (URICA) • Allow users to interactively refine system’s adaptation decision • System learns from user modifications • Uses history for future adaptation predictions • Applicable to a wide range of adaptation types, such as image fidelity and page layout

  7. 40KB Server 1 Improve Fidelity Server 2 Mobile 2 Application 10KB 20KB Prediction How it Works Mobile 1 Application Adaptation Proxy

  8. Predictions based on History

  9. Predictions based on History Challenge: When users have varying preferences, how do we pick an appropriate adaptation?

  10. Varying Adaptation Preferences Leads To “Dirty” History

  11. Correlation-Based Content Adaptation • Typically, web pages/sites contain multiple objects • e.g. images • Use history to determine correlations in the adaptation requirements of different objects • When a user provides corrective feedback for one object, update the adaptation prediction for all related objects!

  12. Feedback!

  13. How To Find Correlations Automatically? • Boosted Decision Stumps • Mine data to create rules that capture relationships between the adaptation requirements of objects For objects X and Y: IF X > 3 THEN Y=7 • Gaussian Mixture Model • History data is used to parameterize a set of Gaussian distributions • Key parameter is # of distributions to consider • A user belongs to each distribution with some prior probability • As the user provides feedback to the system, these probabilities are updated

  14. Page Layout Adaptation Prototype • Intended for use on mobile devices with limited screen real-estate • Allows users to increase or decrease the display size of images on web pages • Key metric is # of user interactions required to reach appropriate adaptation

  15. Page Layout User Study • User study • 3 simulated display sizes: Phone, PDA and in-car browser • 4 web pages, 3 images per page • 30 participants • No prediction during data collection • Traces used to run experiments • Leave-one-out cross validation

  16. Page Layout User Study • User study • 3 simulated display sizes: Phone, PDA and in-car browser • 4 web pages, 3 images per page • 30 participants • No prediction during data collection • Traces used to run experiments • Leave-one-out cross validation Without Correlation History-based predictions: 15 interactions, on average With Correlations Decision Stumps: 5.1 interactions, on average Gaussian Mixture Model: 5.9 interactions, on average

  17. Fidelity Adaptation Prototype • Intended for bandwidth-limited environments

  18. Fidelity Adaptation Prototype • Intended for bandwidth-limited environments

  19. Fidelity Adaptation Prototype • Intended for bandwidth-limited environments

  20. Fidelity Adaptation Prototype • Intended for bandwidth-limited environments

  21. Fidelity Adaptation Prototype (contd.) • Two primary metrics of concern • Number of user interactions • Wasted bandwidth • Users can only increase the fidelity of images • Users have little incentive to reduce the fidelity of an image that they have already been served • Feedback is only one-sided, as opposed to the two-sided feedback received in page layout adaptation

  22. Fidelity Adaptation:Movie Posters Study • User study • Users given 1 of 3 tasks • 9 web pages, 1 image of a movie poster per page • 37 participants per task • No prediction during data collection • Traces used to run experiments • Leave-one-out cross validation

  23. Results From Movie Posters Study GMM (One-sided Feedback)

  24. Results From Movie Posters Study GMM (One-sided Feedback) GMM (Perfect Feedback, Hypothetical)

  25. Results From Movie Posters Study GMM (One-sided Feedback) GMM (Perfect Feedback, Hypothetical) Gaussian Mixture Model and Decision Stumps Did Not Perform Well When Only One-Sided Feedback Is Available

  26. Our Approach • Run standard clustering algorithm (K-Means) on adaptation history • Custom algorithm (called all-in) to perform online classification • Intuition: narrow down the possible clusters a user can belong to quickly

  27. Operation of all-in Algorithm avg, min, max avg, min, max avg, min, max Image Clusters Found In History Using history, precalculate clusters as well as range of fidelities (min,max) for each image

  28. Operation of all-in Algorithm avg, min, max avg, min, max avg, min, max Image Desired By User Provided By System Clusters Found In History When a user initially accesses the page, all clusters are valid

  29. Operation of all-in Algorithm avg, min, max avg, min, max avg, min, max Image Desired By User Provided By System Clusters Found In History For each image, serve the lowest maximum fidelity value from the valid clusters

  30. Operation of all-in Algorithm avg, min, max avg, min, max avg, min, max Image Desired By User Provided By System Clusters Found In History

  31. Operation of all-in Algorithm avg, min, max avg, min, max avg, min, max Image Desired By User Provided By System Clusters Found In History If the user requests a higher fidelity for an image, we can eliminate a cluster

  32. Operation of all-in Algorithm avg, min, max avg, min, max avg, min, max Image Desired By User Provided By System Clusters Found In History If the user requests a higher fidelity for an image, we can eliminate a cluster

  33. Operation of all-in Algorithm avg, min, max avg, min, max avg, min, max Image Desired By User Provided By System Clusters Found In History

  34. Operation of all-in Algorithm avg, min, max avg, min, max avg, min, max Image Desired By User Provided By System Clusters Found In History

  35. Operation of all-in Algorithm avg, min, max avg, min, max avg, min, max Image Desired By User Provided By System Clusters Found In History

  36. Operation of all-in Algorithm avg, min, max avg, min, max avg, min, max Image Desired By User Provided By System Clusters Found In History When only a single cluster remains, serve at the average

  37. Operation of all-in Algorithm avg, min, max avg, min, max avg, min, max Image Desired By User Provided By System Clusters Found In History When only a single cluster remains, serve at the average

  38. Results From Movie Posters Study GMM (One-sided Feedback) GMM (Perfect Feedback, Hypothetical) All-in

  39. Fulfillment Time(Movie Posters Study) Legend: NA: No Adaptation DS: Decision Stumps GM: Gaussian Mixture AI: All-in OR: Hypothetical Oracle Fulfillment Time = download time + time spent by user to provide feedback

  40. Fulfillment Time(Movie Posters Study) Legend: NA: No Adaptation DS: Decision Stumps GM: Gaussian Mixture AI: All-in OR: Hypothetical Oracle Fulfillment Time = download time + time spent by user to provide feedback

  41. Fulfillment Time(Movie Posters Study) Legend: NA: No Adaptation DS: Decision Stumps GM: Gaussian Mixture AI: All-in OR: Hypothetical Oracle Fulfillment Time = download time + time spent by user to provide feedback

  42. Fulfillment Time(Movie Posters Study) Legend: NA: No Adaptation DS: Decision Stumps GM: Gaussian Mixture AI: All-in OR: Hypothetical Oracle Fulfillment Time = download time + time spent by user to provide feedback

  43. Fulfillment Time(Movie Posters Study) Legend: NA: No Adaptation DS: Decision Stumps GM: Gaussian Mixture AI: All-in OR: Hypothetical Oracle Fulfillment Time = download time + time spent by user to provide feedback

  44. Summary • Correlation-based adaptation can be used to provide fine grain customization of content even when users have varying preferences • Standard machine learning techniques work well when there is two-sided feedback (e.g. page layout adaptation) • All-in algorithm performs well when only one-sided feedback is available (e.g. fidelity adaptation) • All-in behaves aggressively to quickly narrow down the number of clusters to which a user can belong

More Related