TrajWiki: Source-Grounded Memory Trajectories for Long-Horizon Dialogue Agents

2026-08-01

AI Summary

Current dialogue agents lack an external memory mechanism that is traceable, updatable, and transparently diagnosable, hindering effective management of information provenance, evolution, and conflicts in long-term conversations. To address this limitation, this work proposes TrajWiki, a novel framework that models memory as source-grounded evolutionary trajectories. It maintains memory consistency through immutable snapshots and declarative operations—ADD, REVISE, and DEPRECATE—and introduces a Memory Wiki intermediate layer to structurally compile dialogue history. This architecture enables hierarchical retrieval and evidence-driven response generation. Empirical results demonstrate that TrajWiki significantly enhances long-context dialogue performance on the LoCoMo and MedMT datasets, benefiting both open- and closed-source large language models while improving the traceability, interpretability, and diagnosability of memory evolution and response generation.

Abstract

Large language model agents have shown strong capabilities in generating coherent and contextually appropriate responses, yet robust long-horizon dialogue remains limited by the lack of external memory that is traceable, updatable, and diagnostically transparent. Existing memory-augmented agents often store memories as isolated records or overwritable states, making it difficult to preserve how information originates, evolves, conflicts, or becomes obsolete over time. We propose TrajWiki, a trajectory-based memory framework for long-horizon conversational agents. Instead of treating memory as static entries, TrajWiki represents each memory as a source-grounded evolution trajectory, maintained through immutable episodic snapshots and claim-level operations such as ADD, REVISE, and DEPRECATE. To reduce fragmentation and retrieval cost, TrajWiki further introduces Memory Wiki, a persistent intermediate layer that incrementally compiles dialogue history into structured and interlinked wiki pages capturing salient entities, events, quantities, topics, and conflicts. At inference time, queries are routed hierarchically from relevant wiki pages to linked memory trajectories, then to corresponding snapshots and source messages for evidence-grounded answer synthesis. Experiments on LoCoMo and MedMT show that TrajWiki improves long-horizon dialogue performance across both open-source and closed-source LLM backbones, while providing greater interpretability and diagnostic visibility into memory evolution, retrieval failures, and answer generation.

Problem

• long-horizon dialogue
• external memory
• memory traceability
• memory evolution
• diagnostic transparency

Innovation

• trajectory-based memory
• source-grounded memory
• Memory Wiki
• long-horizon dialogue
• immutable snapshots