@inproceedings{zhao-etal-2025-motir,
    title = "{M}oti{R}: Motivation-aware Retrieval for Long-Tail Recommendation",
    author = "Zhao, Kaichen  and
      Li, Mingming  and
      Zhao, Haiquan  and
      Liu, Kuien  and
      Li, Zhixu  and
      Li, Xueying",
    editor = "Rehm, Georg  and
      Li, Yunyao",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/mtsummit-25-ingestion/2025.acl-industry.65/",
    doi = "10.18653/v1/2025.acl-industry.65",
    pages = "934--945",
    ISBN = "979-8-89176-288-6",
    abstract = "In the retrieval stage of recommendation systems, two-tower models are widely adopted for their efficiency as a predominant paradigm. However, this method, which relies on collaborative filtering signals, exhibits limitations in modeling similarity for long-tail items. To address this issue, we propose a Motivation-aware Retrieval for Long-Tail Recommendation, named MotiR. The purchase motivations generated by LLMs represent a condensed abstraction of items' intrinsic attributes. By effectively integrating them with traditional item features, this approach enables the two-tower model to capture semantic-level similarities among long-tail items. Furthermore, a gated network-based adaptive weighting mechanism dynamically adjusts representation weights: emphasizing semantic modeling for long-tail items while preserving collaborative signal advantages for popular items. Experimental results demonstrate 60.5{\%} Hit@10 improvements over existing methods on Amazon Books. Industrial deployment in Taobao{\&}Tmall Group 88VIP scenarios achieves over 4{\%} CTR and CVR improvement, validating the effectiveness of our method."
}