Papers by Bingzhe Li
MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents (2026.acl-long)
Copied to clipboard
| Challenge: | Existing approaches rely on semantic similarity over monolithic memory stores, entangling temporal, causal, and entity information. |
| Approach: | They propose a multi-graph agentic memory architecture that represents each memory item across orthogonal semantic, temporal, causal, and entity graphs. |
| Outcome: | Experiments on LoCoMo and LongMemEval show that MAGMA outperforms state-of-the-art models in long-horizon reasoning task. |