UCLA

Recommendation engine is an integral part in digital business nowadays as abundant user interactions could be captured. This paper provides a comprehensive overview on exist- ing approaches to solve implicit feedback recommendation system. It introduces implicit feedback data and the challenge on solving recommendation problem uniquely for implicit feedback. Theoretical aspects sort out the necessary formulation, statistical learning, sampling and optimization approaches. To make theory into practice, several implementations using different combination of statistical learning and optimization methods are applied on real world e-commerce data to illustrate the performance and drawbacks.