Guest Editors
Prof. Didier El Baz
Email: elbaz@laas.fr
Affiliation: LAAS-CNRS, Université de Toulouse, Toulouse, France
Homepage:
Research Interests: optimization, parallel and distributed computing
Dr. Lei Shi
Email: leiky_shi@cuc.edu.cn
Affiliation: State Key Laboratory of Media Convergence and Communication, Communication University of China, China
Homepage:
Research Interests: cross-media retrieval, data analysis and mining, knowledge discovery, artificial intelligence
Dr. Feifei Kou
Email: koufeifei000@bupt.edu.cn
Affiliation: School of Computer Science, Beijing University of Posts and Telecommunications, Beijing, China
Homepage:
Research Interests: artificial intelligence, big data mining, semantic learning, search and recommendation
Dr. Pengfei Zhang
Email: zpf@aust.edu.cn
Affiliation: School of Computer Science and Engineering, Anhui University of Science & Technology, Huainan, China
Homepage:
Research Interests: data privacy and trustworthy artificial intelligence
Summary
With the evolution of recommendation system technologies, both academia and industry are approaching a critical turning point: a transition from competing paradigms toward deep integration and collaboration between traditional discriminative recommendation and emerging generative recommendation.
Traditional recommendation systems, built upon collaborative filtering, deep neural networks, and related methods, have accumulated mature engineering practices and well-established algorithmic frameworks. By leveraging historical behavioral data and modeling user–item interaction patterns, these systems have achieved strong predictive performance. However, they remain constrained by static candidate sets and limited content representation capabilities.
In contrast, generative recommendation systems, empowered by large language models, overcome architectural limitations of traditional frameworks through advanced semantic understanding and content generation. They introduce a new paradigm that enables personalized recommendations generated directly from large-scale corpora. Nevertheless, purely generative approaches face practical challenges, including high inference costs, limited real-time responsiveness, and unstable performance in cold-start scenarios.
This special issue focuses on emerging theories and recent advances in both traditional and generative recommendation technologies. It aims to foster deep collaboration between the two paradigms and to provide theoretical foundations and technical solutions for the development of next-generation, efficient, interpretable, and practically deployable intelligent recommendation systems.
Keywords
generative recommendation, discriminative recommendation, large language model recommendation, retrieval-augmented generation, collaborative filtering, deep neural network recommendation, sequential recommendation, multimodal recommendation, recommender system efficiency optimization, explainable recommendation
Login
Register
Submit a Paper
Propose a Special lssue