Enhanced Service Recommender and Ranking System Using Browsing Patterns of Users

We present an enhanced service recommender system for web services, which takes existing recommendations from services or content providers and based on topics identified by the user's recent browsing history (including past clicks and searches) re-ranks the recommended URLs for each service, giving an aggregate re-ranked recommendation list. This information can be used by third party services for giving more relevant recommendations and notifications to the user, as well as to build a user interests profile. We have implemented our system using a modified version of LDA to cluster the browsing history, and validated it using browsing data gathered from a selection of users.