MARS: Matching Attribute-aware Representations for Text-based Sequential Recommendation

TL;DR

MARS addresses text-based sequential recommendation by representing users and items with multiple attribute-aware representations derived from textual item attributes using a PLM. It introduces attribute-wise interaction matching, where per-attribute scores computed via max similarity are aggregated to form the final score $S(s,i)$, enabling nuanced matching of user interests to item attributes. The approach delivers state-of-the-art performance on five Amazon datasets, with substantial gains in Recall@10 and NDCG@10 and notable zero-shot transfer capabilities, demonstrating effective knowledge transfer from PLMs to unseen domains. By combining Longformer-based attribute encoding with MaxSim-style matching, MARS offers a scalable and interpretable framework for fine-grained, attribute-centric recommendations.

Abstract

Sequential recommendation aims to predict the next item a user is likely to prefer based on their sequential interaction history. Recently, text-based sequential recommendation has emerged as a promising paradigm that uses pre-trained language models to exploit textual item features to enhance performance and facilitate knowledge transfer to unseen datasets. However, existing text-based recommender models still struggle with two key challenges: (i) representing users and items with multiple attributes, and (ii) matching items with complex user interests. To address these challenges, we propose a novel model, Matching Attribute-aware Representations for Text-based Sequential Recommendation (MARS). MARS extracts detailed user and item representations through attribute-aware text encoding, capturing diverse user intents with multiple attribute-aware representations. It then computes user-item scores via attribute-wise interaction matching, effectively capturing attribute-level user preferences. Our extensive experiments demonstrate that MARS significantly outperforms existing sequential models, achieving improvements of up to 24.43% and 29.26% in Recall@10 and NDCG@10 across five benchmark datasets. Code is available at https://github.com/junieberry/MARS

Figures (2)

• Figure 1: A motivating example of two user sequences. The green and pink boxes represent different user interests that are highly relevant to target items.
• Figure 2: Model architecture of MARS. Each item consists of three attributes, e.g., title, brand, and category. For simplicity, we only display attribute values, omitting their keys.