MemGrove — An Autonomous, Tree-Structured Local Memory System for Persistent AI Agents
Fuente:
Zenodo
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Format: | Recurso digital |
| Veröffentlicht: |
Zenodo
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866902279872839680 |
|---|---|
| author | Wang, Zhongren |
| author_facet | Wang, Zhongren |
| contents | <p>Current large language model (LLM)-based dialogue systems lack persistent, personalized long-term memory. While retrieval-augmented generation (RAG) offers short-term context extension, it rarely supports lifelong, user-driven memory accumulation with semantic organization. We present <strong>MemGrove</strong>, an open-source, locally-deployable dual-agent system that endows AI with true long-term memory through a novel <strong>tree-structured memory architecture</strong> managed by an autonomous LLM-powered memory agent. MemGrove enables permanent, private, and self-growing memory storage in human-readable JSON format, supports semantic retrieval via flattened tree indexing, and maintains cross-session continuity without auto-expiry. By shifting memory management from static databases to dynamic, AI-mediated structures, MemGrove offers a practical and privacy-preserving foundation for persistent personal AI agents.</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_17600831 |
| institution | Zenodo |
| language | |
| publishDate | 2025 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | MemGrove — An Autonomous, Tree-Structured Local Memory System for Persistent AI Agents Wang, Zhongren Tree-structured memory Persistent memory MemoryAgent Long-term memory system <p>Current large language model (LLM)-based dialogue systems lack persistent, personalized long-term memory. While retrieval-augmented generation (RAG) offers short-term context extension, it rarely supports lifelong, user-driven memory accumulation with semantic organization. We present <strong>MemGrove</strong>, an open-source, locally-deployable dual-agent system that endows AI with true long-term memory through a novel <strong>tree-structured memory architecture</strong> managed by an autonomous LLM-powered memory agent. MemGrove enables permanent, private, and self-growing memory storage in human-readable JSON format, supports semantic retrieval via flattened tree indexing, and maintains cross-session continuity without auto-expiry. By shifting memory management from static databases to dynamic, AI-mediated structures, MemGrove offers a practical and privacy-preserving foundation for persistent personal AI agents.</p> |
| title | MemGrove — An Autonomous, Tree-Structured Local Memory System for Persistent AI Agents |
| topic | Tree-structured memory Persistent memory MemoryAgent Long-term memory system |
| url | https://doi.org/10.5281/zenodo.17600831 |