中圖分類號： TP18
文獻(xiàn)標(biāo)識碼： A
DOI： 10.19358/j.issn.2096-5133.2021.11.006
引用格式： 洪志理，賴俊，曹雷，等. 融合用戶興趣建模的智能推薦算法研究[J].信息技術(shù)與網(wǎng)絡(luò)安全，2021，40(11)：37-48.

Research on intelligent recommendation algorithm integrating user interest modeling

Abstract： Reinforcement learning is more and more applied to recommendation system. This paper proposes a recommendation method based on DDPG and user dynamic interest modeling(DDPG-LA). It uses LSTM network to extract user′s long-term interest and attention mechanism to extract user′s short-term interest. The two kinds of interest are combined as the state of agent. At the same time, the state enhancement unit is added to LSTM network to accelerate the modeling of users′ long-term interest, and the module to alleviate the recommendation delay is added to the attention mechanism to solve the defects when the method is applied to the recommendation system. In this paper, the model is tested on two data sets of Movelines, and compared with the traditional methods in various test indexes, the results show that the proposed algorithm has more advantages.

Key words : reinforcement learning; recommendation system；DDPG；DDPG-LA；LSTM；attention mechanism；long-term interest；short-term interest