ActionRAG improves long-term memory retrieval for large language model agents
ActionRAG: Evidence-Refining Graph Diffusion for Long-Term Memory in LLM Agents
Artificial IntelligenceComputation and Language
Summary
Large language model (LLM) agents need to remember facts accurately over long conversations, especially when answers depend on how facts connect to each other. The authors identify problems with current memory methods that can mix up or lose important details. To fix this, they created ActionRAG, a system that carefully organizes and refines memory using graphs at query time, improving how well agents retrieve and connect relevant information. Tests show ActionRAG is more accurate and efficient than previous methods.
What this means in practice
- •For chatbot developers: Enhance chatbot memory modules to recover consistent and connected facts from long user interactions using graph-based retrieval refinement.
- •For digital assistant teams: Improve virtual assistants’ ability to accurately recall and relate past interaction details by integrating cautious indexing with graph retroevaluation.
Authors
Navyansh Mahla, Sashank Vandrangi
Abstract
LLM agents need long-term memory that can recover precise and mutually consistent evidence from growing interaction histories, particularly when answers depend on relations among facts rather than isolated semantic matches. Current memory pipelines remain vulnerable at two stages: indexing errors can fragment repeated entities or merge distinct ones, and online retrieval often treats initial similarity scores as final even when the surrounding graph provides evidence for revising them. We introduce ActionRAG, a graph memory architecture that couples cautious indexing with query-time graph retroevaluation. Its indexing pipeline reconciles type labels and applies tiered entity resolution. Given a query, dual-channel entity and triple retrieval forms an initial evidence set and induces a bounded local graph. Query-conditioned receiver-normalized diffusion refines triple scores using local path structure. Seed protection, hub-penalized flux, and triple materialization preserve direct evidence while recovering missed connections. We evaluate ActionRAG on MemoryAgentBench (MAB), which tests accurate retrieval, conflict resolution, long-range understanding, and test-time learning. Across its eight deterministic-scoring columns, ActionRAG achieves 55.1 average accuracy, improving over the strongest retrieval-based memory baseline by 5.4 points. Under MAB's query-time cost accounting, mean synthesis cost stays below \$0.001 per query on the four reported cost columns, and the sum of separately measured retrieval-side p99 components is 440.7 ms on our consumer Apple Silicon setup.