Random Decision Tree; Discriminative pattern extraction; Categorical data clustering, etc. Propose multi-label random decision tree algorithms, which is faster than popular multi-label classification algorithms, moreover the computational cost raises slowly with the increase of the number of labels.
Improve accuracy of collaborative filtering algorithms, multidimensional data based collaborative filtering algorithms, i.e. tag-based recommendation, collaborative filtering parallelization, architecture of recommender system, privacy protection in recommender systems, and social community based recommender systems; design and implementation of personalized recommender systems; and social community based recommender systems.
Research and develop the advertising system for 600 million users based on the data of Tencent.