智枢OSINT:基于深度研究的多智能体情报分析框架OA

Open-source intelligence analysis plays an increasingly important role in supporting complex decision-making.However,traditional intelligence analysis processes are inefficient due to challenges such as the complexity and variability of tasks,the difficulty of processing multi-source heterogeneous information,and the reliance on expert experience for deep reasoning.To address these issues,this paper proposes a deep research agent framework for intelligence analysis.This framework simulates a human intelligence analysis team,constructing a system led by a"Task Planning Agent"and coordinated by specialized sub-agents,including the"Information Collection Agent,""Information Identification Agent,"and"Insight Generation Agent."The framework aims to automate the entire process,from task decomposition to the production of complete intelligence reports.To validate the effectiveness of the framework,we constructed an evaluation dataset for complex intelligence tasks.The results show that the framework significantly outperforms baseline models in terms of generated report quality(RACT),but still has room for improvement in terms of citation performance in web retrieval(FACT).