考虑用户意图和时间间隔的会话型深度学习推荐系统

A SESSION-BASED DEEP LEARNING RECOMMENDATION SYSTEM CONSIDERING USER PURPOSE AND TIME INTERVAL

基于循环神经网络的会话型推荐系统在建模用户点击行为时,无法同时考虑用户行为之间的时间间隔和用户的主要意图。针对该问题,在现有的基于注意力机制的会话型推荐系统和仅考虑用户行为时间间隔的Time-LSTM的深度学习模型的基础上提出一个新的基于会话的推荐系统TASR。利用Time-LSTM建模时间间隔影响用户行为,并利用注意力机制捕获用户的主要意图。在两个公开数据集上的实验验证了该算法的有效性。

The session-based recommendation systems with recurrent neural networks cannot simultaneously consider the time interval between user behaviors and the user’s main purpose when modeling user click behavior.Based on the existing session-based recommendation system with attention mechanism and the deep learning model of Time-LSTM considering only the time interval of user behavior,a new session-based recommendation system named TASR is proposed in this paper.It used Time-LSTM to model time intervals to refelt user behavior and used the attention mechanism to capture the user s main intent.The experiments on two published data sets verify the effectiveness of the algorithm.