Web Page Recommender System

The internship involves research into technology that can recommend web sites to a user. The user indicates web sites they like, and the recommender engine suggests pages with related content, or which people similar to the user liked. A recommender technology has been in development at Worio, a search engine company. What is proposed here are three machine learning projects that extend the recommender technology. First, learning a more sophisticated model of each user, by looking at all activity a user does, rather than just indicating preference. Second, selecting pages to recommend which maximizes what the system learns about the user (for instance, adding a few popular political sites to see if they visit them). Third, developing an interface that allows users to “train” the system by giving more direct and detailed feedback, so that, for example, they have the option saying they generally prefer political comedy sites to other comedy sites.

Faculty Supervisor:

Dr. Nando de Freitas

Student:

Eric Brochu

Partner:

Worio

Discipline:

Computer science

Sector:

Information and communications technologies

University:

University of British Columbia