Tag
This paper introduces a neuro-symbolic meta-policy for temporal knowledge-graph memory in partially observable reinforcement learning, combining RDF-based graph representations with symbolic memory management heuristics to achieve inspectable and adaptive control.
The article argues that the default approach to agent memory (saving all messages and using RAG) fails due to lack of suppression, identity, and relation tracking, and proposes using a lightweight temporal knowledge graph to model the user's world instead.
A detailed architectural guide for building long-running AI agents that handle changing user preferences over time by combining a vector store, graph DB, and temporal edges instead of overwriting data.