Jianghao Lin,Jiaqi Liu,Jiachen Zhu,Yunjia Xi,Chengkai Liu,Yangtian Zhang,Yong Yu,Weinan Zhang
Abstract:While traditional recommendation techniques have made significant strides in the past decades, they still suffer from limited generalization performance caused by factors like inadequate collaborative signals, weak latent representations, and noisy data. In response, diffusion models (DMs) have emerged as promising solutions for recommender systems due to their robust generative capabilities, solid theoretical foundations, and improved training stability. To this end, in this paper, we present the first comprehensive survey on diffusion models for recommendation, and draw a bird's-eye view from the perspective of the whole pipeline in real-world recommender systems. We systematically categorize existing research works into three primary domains: (1) diffusion for data engineering & encoding, focusing on data augmentation and representation enhancement; (2) diffusion as recommender models, employing diffusion models to directly estimate user preferences and rank items; and (3) diffusion for content presentation, utilizing diffusion models to generate personalized content such as fashion and advertisement creatives. Our taxonomy highlights the unique strengths of diffusion models in capturing complex data distributions and generating high-quality, diverse samples that closely align with user preferences. We also summarize the core characteristics of the adapting diffusion models for recommendation, and further identify key areas for future exploration, which helps establish a roadmap for researchers and practitioners seeking to advance recommender systems through the innovative application of diffusion models. To further facilitate the research community of recommender systems based on diffusion models, we actively maintain a GitHub repository for papers and other related resources in this rising direction <a class="link-external link-https" href="https://github.com/CHIANGEL/Awesome-Diffusion-for-RecSys" rel="external noopener nofollow">this https URL</a>.
What problem does this paper attempt to address?
### Problems the Paper Aims to Solve
The paper "A Survey on Diffusion Models in Recommender Systems" aims to address the following issues:
1. **Information Overload Problem**: With the rapid development of online services, recommender systems (RSs) have become key tools for alleviating information overload and providing personalized content. However, traditional recommendation techniques still have limitations in performance, such as insufficient collaborative signals, weak latent representations, and data noise.
2. **Improving Recommender System Performance**: Diffusion models (DMs), known for their excellent generative capabilities and representation learning abilities, are seen as promising solutions to enhance recommender system performance. The paper aims to systematically review the application of diffusion models in recommender systems, showcasing their unique advantages in capturing complex data distributions and generating high-quality, diverse samples.
3. **Establishing a Research Roadmap**: Although diffusion models have achieved significant success in other fields (such as computer vision, audio generation, natural language processing, and reinforcement learning), their application in recommender systems is still in its early stages. The paper aims to summarize existing research progress, identify key challenges, and propose future research directions, providing researchers and practitioners with a clear technical roadmap.
### Main Contributions
1. **Comprehensive and Up-to-date Survey**: This is the first comprehensive, up-to-date, and forward-looking survey on the application of diffusion models in recommender systems. The paper highlights the applicability of diffusion models in recommender systems and discusses their advantages from multiple perspectives.
2. **Unified and Structured Classification**: The paper introduces a well-organized classification method, dividing existing research work into three categories: diffusion models in data engineering and encoding, diffusion models as recommendation models, and diffusion models for content presentation. This classification provides readers with a coherent roadmap, helping to understand the application trends of diffusion models in recommender systems from multiple angles.
3. **Insights into Challenges and Future Directions**: The paper identifies key challenges in the current research field and further points out several promising future research directions, aiming to attract more researchers to the study of diffusion model-based recommender systems.
Through these contributions, the paper not only provides a systematic summary of existing research but also offers valuable guidance for future exploration.