CAM: A Constructivist View of Agentic Memory for LLM-Based Reading Comprehension

Xu Chen (Renmin University of China) · Ruiming Tang (Kuaishou- 快手科技) · Rui Li (Rochester Institute of Technology) · Zeyu Zhang (The Australian National University) · Xiaohe Bo (Renmin University of China) · Zihang Tian (Renmin University of China) · Quanyu Dai (Huawei Technologies Ltd.) · Zhenhua Dong (Huawei Technologies Ltd.)
agentic memoryassociative processautonomous reading agentscamclaim verificationcoherent hierarchical summarizationcohesive memory moduleconstructivist theorycontextual responsedynamic accommodationflexible assimilationincremental overlapping clustering algorithmonline batch integrationreading comprehension tasksstructured schemata

Current Large Language Models (LLMs) are confronted with overwhelming information volume when comprehending long-form documents. This challenge raises the imperative of a cohesive memory module, which can elevate vanilla LLMs into autonomous reading agents. Despite the emergence of some heuristic approaches, a systematic design principle remains absent. To fill this void, we draw inspiration from Jean Piaget's Constructivist Theory, illuminating three traits of the agentic memory