Agent4Rec

Agent4Rec

Open Source

Agent4Rec is a recommender system simulator that utilizes 1,000 LLM-empowered generative agents. These agents, initialized from the MovieLens-1M dataset, exhibit diverse social traits and preferences.

Agent4Rec is a project that showcases the implementation of generative agents in recommendation systems, contributing innovative perspectives for those interested in this technology. This implementation is particularly relevant for the SIGIR 2024 conference, where cutting-edge research in the field will be highlighted. The project's open-source nature allows developers and researchers to explore and adapt the techniques presented.

Agent4Rec Key Features

  • Allows for customizable simulation settings
  • Records interaction history of agents
  • Simulates 1,000 LLM-empowered generative agents
  • Supports various recommendation algorithms (Random, Pop, MF, MultVAE, LightGCN)
  • Utilizes the MovieLens-1M dataset

Agent4Rec Use Cases

  • Exploring user behavior modeling in recommender systems
  • Research on generative agents in recommendation contexts
  • Simulating user interactions in recommendation systems
  • Testing and evaluating recommendation algorithms