Elamur Research⌕
☰
home / papers / 2026 / Inside Out: Evolving User-Centric Core Memory Trees for Long-Term Personalized Dialogue Systems

Inside Out: Evolving User-Centric Core Memory Trees for Long-Term Personalized Dialogue Systems

Paper · 2026 · Jihao Zhao, Ding Chen, Zhaoxin Fan et al.

Summary

Personalized dialogue agents that talk with a user over a long period run into a conflict between an ever-growing history and a fixed context window. Feeding the whole history to the model adds noise, raises cost, weakens reasoning and makes the persona drift. Inside Out addresses this by keeping a single global user profile, the PersonaTree. Its trunk is fixed by a schema derived from the Biopsychosocial model, while its branches and leaves are edited as dialogue chunks arrive.

The edits come from MemListener, a small model trained first with SFT and then with reinforcement learning using process-based rewards. It emits atomic, interpretable ADD, UPDATE, DELETE and NO_OP operations. When answering, the system has two modes:

  • a fast mode that conditions directly on the tree;
  • an agentic mode that retrieves extra details from the history, guided by the tree.

Experiments show PersonaTree beats full-history concatenation and several memory systems while using far less context. MemListener matches or exceeds large reasoning models at memory-operation decisions. This points to a low-cost setup in which a small model maintains memory and a large LLM generates responses.

Key points

  • Training data and setup. About 28k instructions were built for training, including PersonaMem 15K and HaluMem 13K. In the RL setting, HaluMem 13K serves for SFT warm-up and 0.5K PersonaMem samples for process-reward training.
  • Context efficiency. The ALL Dialogue baseline needs roughly 32K tokens of context. The trained PersonaTree adds only about 2.2K–2.6K tokens of memory, yet matches or beats that baseline with Longcat-Flash-Chat and DeepSeek-R1-0528 as response models.
  • Comparison with strong models. MemListener's memory-operation decisions are reported as comparable to, or better than, those of DeepSeek-R1-0528 and Gemini-3-Pro.
  • Operation-generation strategy. Generating tree operations directly outperforms an extract-then-transform pipeline in every model combination tested, by about +0.34 to +3.06.
  • Chunk size and code. A chunk window of w = 3 dialogue turns gave the most stable results, for both DeepSeek-R1-0528 and Qwen2.5-7B-RL as tree-extraction models. Code is available at github.com/MemTensor/InsideOut.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 works

🧭 Topics 1Research subjects and languages the work is about.Dialogue systems · 654 works

Cite

Cite as Jihao Zhao, Ding Chen, Zhaoxin Fan, Kerun Xu, Mengting Hu, Bo Tang, Feiyu Xiong and Zhiyu li. 2026. Inside Out: Evolving User-Centric Core Memory Trees for Long-Term Personalized Dialogue Systems. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.614

@inproceedings{zhao2026inside,
  title = {{Inside Out: Evolving User-Centric Core Memory Trees for Long-Term Personalized Dialogue Systems}},
  author = {Jihao Zhao, Ding Chen, Zhaoxin Fan, Kerun Xu, Mengting Hu, Bo Tang, Feiyu Xiong and Zhiyu li},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.614},
  url = {https://doi.org/10.18653/v1/2026.acl-long.614}
}

References

29 references, 11 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Aili Chen, Chengyu Du, Jiangjie Chen, Jinghan Xu, Yikai Zhang, Siyu Yuan, Zulong Chen, Liangyue Li, and Yanghua Xiao. 2025a. Deeper insight into your user: Directed persona refinement for dynamic persona modeling. arXiv preprint arXiv:2502.11078. find
  2. Ding Chen, Simin Niu, Kehang Li, Peng Liu, Xiangping Zheng, Bo Tang, Xinchi Li, Feiyu Xiong, and Zhiyu Li. 2025b. Halumem: Evaluating hallucinations in memory systems of agents. arXiv preprint arXiv:2511.03506. — [arXiv:2511.03506](https://arxiv.org/abs/2511.03506) find
  3. Yi-Pei Chen, Noriki Nishida, Hideki Nakayama, and Yuji Matsumoto. 2024. Recent trends in personalized dialogue generation: A review of datasets, methodologies, and evaluations. arXiv preprint arXiv:2405.17974. find
  4. Prateek Chhikara, Dev Khant, Saket Aryan, Taranjeet Singh, and Deshraj Yadav. 2025. Mem0: Building production-ready ai agents with scalable long-term memory. arXiv preprint arXiv:2504.19413.
  5. Xu Han, Bin Guo, Yoon Jung, Benjamin Yao, Yu Zhang, Xiaohu Liu, and Chenlei Guo. 2023. Person-apkt: Building personalized dialogue agents via parameter-efficient knowledge transfer. arXiv preprint arXiv:2306.08126. — [arXiv:2306.08126](https://arxiv.org/abs/2306.08126) find
  6. Qiushi Huang, Shuai Fu, Xubo Liu, Wenwu Wang, Tom Ko, Yu Zhang, and Lilian Tang. 2023. Learning retrieval augmentation for personalized dialogue generation. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 2523–2540. find
  7. Bowen Jiang, Zhuoqun Hao, Young-Min Cho, Bryan Li, Yuan Yuan, Sihao Chen, Lyle Ungar, Camillo J Taylor, and Dan Roth. 2025. Know me, respond to me: Benchmarking llms for dynamic user profiling and personalized responses at scale. arXiv preprint arXiv:2504.14225. — [arXiv:2504.14225](https://arxiv.org/abs/2504.14225) find
  8. Jiazheng Kang, Mingming Ji, Zhe Zhao, and Ting Bai. 2025. Memory os of ai agent. arXiv preprint arXiv:2506.06326.
  9. Guanrong Li, Xinyu Liu, Zhen Wu, and Xinyu Dai. 2025a. Persona-aware alignment framework for personalized dialogue generation. Transactions of the Association for Computational Linguistics, 13:1722– 1742. find
  10. Jiwei Li, Michel Galley, Chris Brockett, Georgios Spithourakis, Jianfeng Gao, and William B Dolan. 2016. A persona-based neural conversation model. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 994–1003.
  11. Zhiyu Li, Shichao Song, Chenyang Xi, Hanyu Wang, Chen Tang, Simin Niu, Ding Chen, Jiawei Yang, Chunyu Li, Qingchen Yu, Jihao Zhao, Yezhaohui Wang, Peng Liu, Zehao Lin, Pengyuan Wang, Jiahao Huo, Tianyi Chen, Kai Chen, Kehang Li, and 20 others. 2025b. Memos: A memory os for ai system. arXiv preprint… — [arXiv:2507.03724](https://arxiv.org/abs/2507.03724) find
  12. Lei Liu, Xiaoyan Yang, Yue Shen, Binbin Hu, Zhiqiang Zhang, Jinjie Gu, and Guannan Zhang. 2023. Think-in-memory: Recalling and post-thinking enable llms with long-term memory. Preprint, arXiv:2311.08719. — [arXiv:2311.08719](https://arxiv.org/abs/2311.08719) find
  13. Nelson F. Liu, Kevin Lin, John Hewitt, Ashwin Paranjape, Michele Bevilacqua, Fabio Petroni, and Percy Liang. 2024. Lost in the middle: How language models use long contexts. Transactions of the Association for Computational Linguistics, 12:157–173.
  14. Andrea Madotto, Zhaojiang Lin, Chien-Sheng Wu, and Pascale Fung. 2019. Personalizing dialogue agents via meta-learning. In Proceedings of the 57th annual meeting of the association for computational linguistics, pages 5454–5459.
  15. Hazel Rose Markus. 1977. Self-schemata and processing information about the self. Journal of Personality and Social Psychology, 35:63–78. find
  16. Kai Tzu-iunn Ong, Namyoung Kim, Minju Gwak, Hyungjoo Chae, Taeyoon Kwon, Yohan Jo, Seung-won Hwang, Dongha Lee, and Jinyoung Yeo. 2025. Towards lifelong dialogue agents via timeline-based memory management. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Associati… find
  17. Atsushi Otsuka, Kazuya Matsuo, Ryo Ishii, Narichika Nomoto, and Hiroaki Sugiyama. 2024. User-specific dialogue generation with user profile-aware pre-training model and parameter-efficient fine-tuning. arXiv preprint arXiv:2409.00887. — [arXiv:2409.00887](https://arxiv.org/abs/2409.00887) find
  18. Preston Rasmussen, Pavlo Paliychuk, Travis Beauvais, Jack Ryan, and Daniel Chalef. 2025. Zep: A temporal knowledge graph architecture for agent memory. Preprint, arXiv:2501.13956.
  19. Alireza Salemi, Sheshera Mysore, Michael Bendersky, and Hamed Zamani. 2024. LaMP: When large language models meet personalization. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 7370–7392, Bangkok, Thailand. Association for C…
  20. Zhaoxuan Tan and Meng Jiang. 2023. User modeling in the era of large language models: Current research and future directions. Preprint, arXiv:2312.11518. — [arXiv:2312.11518](https://arxiv.org/abs/2312.11518) find
  21. Zhaoxuan Tan, Qingkai Zeng, Yijun Tian, Zheyuan Liu, Bing Yin, and Meng Jiang. 2024. Democratizing large language models via personalized parameter-efficient fine-tuning. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 6476–6491, Miami, Florida, USA.…
  22. Piiastiina Tikka and Harri Oinas-Kukkonen. 2019. Tailoring persuasive technology: A systematic review of literature of self-schema theory and transformative learning theory in persuasive technology context. find
  23. Guangxuan Xiao, Yuandong Tian, Beidi Chen, Song Han, and Mike Lewis. 2024. Efficient streaming language models with attention sinks. In The Twelfth International Conference on Learning Representations. find
  24. Wujiang Xu, Zujie Liang, Kai Mei, Hang Gao, Juntao Tan, and Yongfeng Zhang. 2025. A-mem: Agentic memory for llm agents. arXiv preprint arXiv:2502.12110.
  25. Ori Yoran, Tomer Wolfson, Ori Ram, and Jonathan Berant. 2024. Making retrieval-augmented language models robust to irrelevant context. In ICLR 2024 Workshop on Large Language Model (LLM) Agents.
  26. Saizheng Zhang, Emily Dinan, Jack Urbanek, Arthur Szlam, Douwe Kiela, and Jason Weston. 2018. Personalizing dialogue agents: I have a dog, do you have pets too? arXiv preprint arXiv:1801.07243.
  27. Yuze Zhao, Jintao Huang, Jinghan Hu, Xingjun Wang, Yunlin Mao, Daoze Zhang, Zeyinzi Jiang, Zhikai Wu, Baole Ai, Ang Wang, Wenmeng Zhou, and Yingda Chen. 2025. Swift: a scalable lightweight infrastructure for fine-tuning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pa… find
  28. Yinhe Zheng, Guanyi Chen, Minlie Huang, Song Liu, and Xuan Zhu. 2019. Personalized dialogue generation with diversified traits. arXiv preprint arXiv:1901.09672. — [arXiv:1901.09672](https://arxiv.org/abs/1901.09672) find
  29. Wanjun Zhong, Lianghong Guo, Qiqi Gao, He Ye, and Yanlin Wang. 2024. Memorybank: enhancing large language models with long-term memory. In Proceedings of the Thirty-Eighth AAAI Conference on Artificial Intelligence and Thirty-Sixth Conference on Innovative Applications of Artificial Intelligence and… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhao-2026-inside-out-evolving-user-centric/