Recommendation Explorer

Recommendation Explorer is a recommendation system I'm working on with Miles Efron. It uses knowledge discovery techniques to improve its representation of item-item relationships, and provides a graphical user interface to enable users to explore recommendations in the context of their information needs. Recommendation Explorer currently uses a database of 12,726 records to recommend films, but is being designed as a generic system that can be readily adapted to a variety of resource collections.

The goal of Recommendation Explorer is to minimize user effort while producing high quality, personalized recommendations.  Our approach is two-pronged: First, we use knowledge discovery techniques to develop an effective recommendation engine that requires minimal user input. Our current research module uses dimensionality reduction techniques to discover high-order relationships between resources, thereby producing quality recommendations based on sparse data.

Second, we are creating an interface that enables the user to explore, manipulate, and preview recommended resources. Stored, modifiable profiles and a small number of simple widgets enable the user to provide input to the system quickly and easily and thus personalize the recommendations to his or her immediate context.

More Information

• Recommendation Explorer project Web site
• Efron, Miles and Geisler, Gary (2001). "Is it all About Connections? Factors Affecting the Performance of a Link-Based Recommender System." Proceedings of the SIGIR 2001 Workshop on Recommender Systems (September 13, 2001, New Orleans, LA) [ ].
• Efron, Miles and Geisler, Gary (2001). "Using Dimensionality Reduction to Improve Similarity Judgements for Recommendation." Proceedings of the Second DELOS Network of Excellence Workshop on Personalisation and Recommender Systems in Digital Libraries (June 18-20, 2001, Dublin, Ireland). [  ].