Nowadays, sequential recommender systems are widely used in E-commerce fields to capture consumers’ dynamic preferences in short terms. Existing transformer-based recommendation models mainly consider consumer preference for the products and some related features, such as price. However, besides such objective features, some subjective features, such as consumers’ preference for product quality, also affect consumers’ purchase decisions. In this paper, we design a Sequential Recommender system based on Objective and Subjective features (SROS). We construct subjective features by using natural language processing to analyze online consumer reviews. Then we design a feature-level multi-head self-attention to explore the interactions between objective features and subjective features and capture consumers’ dynamic preferences for them among different purchases. Experimental results on real-world datasets demonstrate the effectiveness of the proposed model.


Paper Number 1469; Track AI; Short Paper



When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.