Hybrid Recommendation Algorithm Based on Long-term and Short-term Interest and Matrix Factorization for Collaborative Filtering
Yong Kang Xu, Ni Zhu · Journal of Physics Conference Series · 2020
This paper proposes a hybrid recommendation algorithm which based on user interests and matrix factorization. It solves problems such as spares data, the difference between long-term and short-term interest, overemphasis of time and ignoring user new interests. The proposed algorithm distinguished users' interest through time window. Then it obtained the distribution of user interests. Finally, it integrated into matrix factorization to explore more new interests. The result shows that short-term interest should be recommend first. Compared with the traditional matrix factorization for collaborative filtering, forgetting curve and time window, the proposed algorithm shows superior performance on precision, recall, and F1-Score. It extracts user's interest sets, user activity, item popularity and other related indicators, which can automatically tag users and provide more options for subsequent research on the dynamic evolution of user interests or the expansion of website functions.