Improving Session-Based Recommendation in E-commerce with FAPAT
Session-based recommendation in e-commerce aims to accurately predict the next item an anonymous user will purchase using their browsing and purchase history. However, existing methods that rely on global or local transition graphs to supplement session data can introduce noisy correlations and obscure user intent. In this article, we introduce the Frequent Attribute Pattern Augmented Transformer (FAPAT) approach, which effectively characterizes user intents by constructing attribute transition graphs and matching attribute patterns.
FAPAT leverages frequent and compact attribute patterns as memory to enhance session representations. This is achieved through a gate and a transformer block that fuse the entire session information. To validate the effectiveness of FAPAT, we conducted extensive experiments on two public benchmarks and analyzed 100 million industrial data across three domains.
The results showcase the superiority of FAPAT over state-of-the-art methods, with an average improvement of 4.5% across various evaluation metrics including Hits, NDCG, and MRR. Furthermore, FAPAT not only improves next-item prediction accuracy but also demonstrates its capabilities to capture user intents by accurately predicting items’ attributes and offering period-item recommendations.
Abstract:The goal of session-based recommendation in E-commerce is to predict the next item that an anonymous user will purchase based on the browsing and purchase history. However, constructing global or local transition graphs to supplement session data can lead to noisy correlations and user intent vanishing. In this work, we propose the Frequent Attribute Pattern Augmented Transformer (FAPAT) that characterizes user intents by building attribute transition graphs and matching attribute patterns. Specifically, the frequent and compact attribute patterns are served as memory to augment session representations, followed by a gate and a transformer block to fuse the whole session information. Through extensive experiments on two public benchmarks and 100 million industrial data in three domains, we demonstrate that FAPAT consistently outperforms state-of-the-art methods by an average of 4.5% across various evaluation metrics (Hits, NDCG, MRR). Besides evaluating the next-item prediction, we estimate the models’ capabilities to capture user intents via predicting items’ attributes and period-item recommendations.