Xinzhuo Wu, Weihai Lu, Zhejun Zhao · ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 2026
DOI: 10.1145/3799682.3840965
Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).
Cross-Domain Sequential Recommendation (CDSR) has emerged as a promising paradigm for alleviating data sparsity by transferring user preferences across domains. However, existing methods typically rely on sparse overlapping users or static LLM-derived embeddings, leading to two challenges: the Overlap Dilemma, where fragmented cross-domain connectivity impedes effective knowledge transfer, and the Semantic Gap, where identical items lack domain-specific contextual interpretations. To address these challenges, we propose Domain-Aware Hierarchical Semantic Representation (DA-HSR), a unified framework for learning transferable yet domain-sensitive representations. Specifically, we develop a Hierarchical Semantic Representation module that employs shared codebooks with domain-adaptive codeword refinements through Domain-adaptive Residual Quantization, thereby disentangling global static knowledge from layered dynamic semantics. We further design a Sequence-aware Routing Fusion module that treats the output of each codebook layer as an independent representation component and applies sequence-conditioned cross-attention to select the most relevant semantic granularity for each user context. To ensure robust cross-domain transfer, we introduce Cross-domain Sequence Encoding to retrieve informative global histories, together with Denoising Cross-domain Interest Transfer to suppress negative noise during preference alignment. Experiments demonstrate that DA-HSR significantly outperforms state-of-the-art baselines while effectively bridging both the semantic and data gaps in CDSR. Code is available at https://github.com/zaixiaxiaowu/DA-HSR.
ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 0 citations
ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 0 citations
ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 0 citations
ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 0 citations
ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 0 citations
No comments yet — start the discussion below.