Multi Interests for Joint Search-Recommendation Modeling

2026-08-11

AI Summary

Existing cross-domain unified modeling approaches struggle to capture fine-grained interest discrepancies across different scenarios within mixed search and recommendation sequences. To address this limitation, this work proposes the MIJSR framework, which jointly models users’ multiple interests from both structural and semantic perspectives: structurally through subsequence partitioning and masking, and semantically via query clustering and semantic segmentation. The framework further employs contrastive learning and a progressive hierarchical multi-task prediction mechanism to enable adaptive fusion of heterogeneous signals. Evaluated on two public datasets, MIJSR significantly improves accuracy for both search and recommendation tasks, demonstrating the effectiveness and superiority of fine-grained multi-interest modeling in cross-domain sequential behavior understanding.

Abstract

Search and recommendation are crucial for understanding user preferences. More and more studies are attempting to jointly model search behavior and recommendation behavior, by integrating user active search and passive recommendation behavior data to better mine user preferences. However, although existing cross-domain unified modeling frameworks can effectively compensate for the differences in behavior between domains, they overlook the expression of interests in different scenarios under mixed sequences. In this study, we propose a multi-interest-based mixed sequential modeling framework MIJSR, which performs multi-interest mining and adaptive integration on search recommendation mixed sequences from both structural and semantic perspectives. Specifically, our model can be roughly divided into three modules: cross-domain behavior fusion, multi-interest mining, and multi-task prediction. Firstly, we align the representations of query and item through contrastive learning training. Then, we extract the multi interests of the mixed behavior sequence from both structural and semantic perspectives. Structurally, we extract search interests, recommendation interests, and cross interests through subsequence partitioning and mask settings; In terms of semantics, we use the semantic information of queries for clustering and perform semantic segmentation on mixed sequences to construct semantic multi interests. Finally, the adaptive fusion of multiple interests is combined with other side information to use a progressive layered extraction model for multi-task prediction. Extensive experiments on two open-source datasets have shown that our model can further enhance its accuracy in search and recommendation by extracting users' multi interests at a fine-grained level. Codes are available at https://github.com/pxcstart/MIJSR.

Problem

• multi-interest
• search-recommendation
• mixed sequences
• cross-domain modeling
• user preferences

Innovation

• multi-interest modeling
• joint search-recommendation
• mixed sequential behavior
• contrastive learning
• semantic clustering