Dynamic context management in context-aware recommender systems. (April 2023)

• Record Type: Journal Article
• Authors: Ali, Waqar; Kumar, Jay; Mawuli, Cobbinah Bernard; She, Lei; Shao, Jie
• Is Part Of: Computers & electrical engineering. Volume 107(2023)
• Publication Date: 2023-04
• Subjects: Context-aware recommender system -- Evolving context behavior -- Dynamic context modeling -- Rating prediction -- Reliable recommendations; 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.2023.108622
• Languages: English
• ISSNs: 0045-7906

Abstract

Abstract: Context-aware recommendation is an essential part of advanced advertising systems. Most of the existing context-aware recommender systems (CARS) build recommendation models considering context as a static entity. In practice, user preferences are dynamic and change over time. In this study, we argue that CARS must be able to dynamically opt for the evolving behavior of contextual attributes. We propose a solution with a runtime identification of contextual information and jointly integrate content analysis to boost recommendation performance. Initially, the proposed model constructs a list of contextually similar candidates. The users' context information extracted in the initial step is modeled together with the content information to estimate a context score used to find contextually reliable neighbors. Instead of relying on a fixed rating matrix, we dynamically exploit contextual information for effective rating prediction. Experiments on three real-world datasets demonstrate that our method outperforms classical and state-of-the-art recommendation algorithms. Graphical abstract: Highlights: A novel technique captures evolving behavior of contextual attributes for context-aware recommendation. Dynamic context management utilizes a modified form of the Minkowski distance for candidate generation. Advantageous for highly sparse e-commerce applications, especially for streaming environments. Evaluation on three diverse datasets highlights the significance of the proposed method.