Context-aware session recommendation based on recurrent neural networks. (May 2022)

• Record Type: Journal Article
• Authors: Wu, Tianhui; Sun, Fuzhen; Dong, Jiawei; Wang, Zhen; Li, Yan
• Is Part Of: Computers & electrical engineering. Volume 100(2022)
• Publication Date: 2022-05
• Subjects: Session-based recommendation -- Contextual recommendation -- Recurrent neural networks -- User preference -- The gated recurrent unit; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2022.107916
• Languages: English
• ISSNs: 0045-7906

Abstract

Abstract: A session-based recommendation system that helps users get the information they are interested in is an important category of personalized recommendation systems. Traditionally, session recommendation algorithms do not take full advantage of users' contextual information. It becomes easier to get users' preferences and context with the rapid development of mobile devices. Under such circumstances, we proposed a novel recommendation algorithm joined session-based context-aware recommendation model. The model maps contextual information into low-dimensional real vector features and then fuses them into a recurrent neural network recommendation model based on sessions by three combinations of Add, Stack, and Multilayer Perceptron. We have verified its extensibility by combining it with the functional extension module which rest on long sequences. We conducted extensive experiments on two public datasets. The experimental results show that our model significantly outperforms state-of-the-art recommendation models in terms of recommendation performance. Graphical abstract: Highlights: Embedding the session information to a low-dimensional RNN with hidden states of gated recurrent units. Low-dimensional features are integrated through a novel combination of Add, Stack, and Multilayer Perceptron. The proposed model is combined with a function extension model for long-term session recommendations.