[공지사항]을 빙자한 안부와 근황 
Show more

Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

DUALRec: A Hybrid Sequential and Language Model Framework for Context-Aware Movie Recommendation

Created by
  • Haebom

Author

Yitong Li, Raoul Grasman

Outline

In this paper, we propose DUALRec to address the challenge of modern recommender systems, which struggle to model and predict users’ dynamic and context-rich preferences. DUALRec combines the temporal modeling capabilities of LSTM networks with the semantic reasoning capabilities of a fine-tuned large-scale language model (LLM) to capture users’ evolving preferences and recommend the next movie. The LSTM captures users’ evolving preferences from their viewing history, and the fine-tuned LLM leverages these temporal user insights to generate the next movie that the users may enjoy. Experimental results on the MovieLens-1M dataset show that the DUALRec model outperforms various baseline models.

Takeaways, Limitations

•
Takeaways:
◦
We present a novel recommender system architecture that effectively models users' dynamic preferences by combining LSTM and fine-tuned LLM.
◦
Combining temporal sequential modeling and semantic reasoning presents a promising direction for developing more intelligent and context-aware recommender systems.
◦
It demonstrates better performance than existing models in various evaluation metrics (HR@k, NDCG@k, genre similarity) on the MovieLens-1M dataset.
•
Limitations:
◦
Since we conducted our experiments using only one dataset, MovieLens-1M, further research is needed to determine generalizability to other datasets.
◦
There is a lack of analysis of the complexity and computational cost of the proposed model.
◦
There may be a lack of detailed explanations of LLM's fine-tuning methods and parameters.
👍
Made with Slashpage