SGMem: Sentence Graph Memory for Long-Term Conversational Agents
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915513826803712 |
|---|---|
| author | Wu, Yaxiong Zhang, Yongyue Liang, Sheng Liu, Yong |
| author_facet | Wu, Yaxiong Zhang, Yongyue Liang, Sheng Liu, Yong |
| contents | Long-term conversational agents require effective memory management to handle dialogue histories that exceed the context window of large language models (LLMs). Existing methods based on fact extraction or summarization reduce redundancy but struggle to organize and retrieve relevant information across different granularities of dialogue and generated memory. We introduce SGMem (Sentence Graph Memory), which represents dialogue as sentence-level graphs within chunked units, capturing associations across turn-, round-, and session-level contexts. By combining retrieved raw dialogue with generated memory such as summaries, facts and insights, SGMem supplies LLMs with coherent and relevant context for response generation. Experiments on LongMemEval and LoCoMo show that SGMem consistently improves accuracy and outperforms strong baselines in long-term conversational question answering. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_21212 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SGMem: Sentence Graph Memory for Long-Term Conversational Agents Wu, Yaxiong Zhang, Yongyue Liang, Sheng Liu, Yong Computation and Language Information Retrieval I.2.7; H.3.3 Long-term conversational agents require effective memory management to handle dialogue histories that exceed the context window of large language models (LLMs). Existing methods based on fact extraction or summarization reduce redundancy but struggle to organize and retrieve relevant information across different granularities of dialogue and generated memory. We introduce SGMem (Sentence Graph Memory), which represents dialogue as sentence-level graphs within chunked units, capturing associations across turn-, round-, and session-level contexts. By combining retrieved raw dialogue with generated memory such as summaries, facts and insights, SGMem supplies LLMs with coherent and relevant context for response generation. Experiments on LongMemEval and LoCoMo show that SGMem consistently improves accuracy and outperforms strong baselines in long-term conversational question answering. |
| title | SGMem: Sentence Graph Memory for Long-Term Conversational Agents |
| topic | Computation and Language Information Retrieval I.2.7; H.3.3 |
| url | https://arxiv.org/abs/2509.21212 |