cs.CLApr 9, 2026

HyperMem: Hypergraph Memory for Long-Term Conversations

Authors: Juwei Yue, Chuanrui Hu, Jiawei Sheng, Zuyi Zhou, Wenyuan Zhang, Tingwen Liu, Li Guo, Yafeng Deng

Organizations: Institute of Information Engineering, Chinese Academy of Sciences · School of Cyber Security, University of Chinese Academy of Sciences · EverMind AI

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Long-term memory is essential for conversational agents to maintain coherence, track persistent tasks, and provide personalized interactions across extended dialogues. However, existing approaches as Retrieval-Augmented Generation (RAG) and graph-based memory mostly rely on pairwise relations, which can hardly capture high-order associations, i.e., joint dependencies among multiple elements, causing fragmented retrieval. To this end, we propose HyperMem, a hypergraph-based hierarchical memory architecture that explicitly models such associations using hyperedges. Particularly, HyperMem structures memory into three levels: topics, episodes, and facts, and groups related episodes and their facts via hyperedges, unifying scattered content into coherent units. Leveraging this structure, we design a hybrid lexical-semantic index and a coarse-to-fine retrieval strategy, supporting accurate and efficient retrieval of high-order associations. Experiments on the LoCoMo benchmark show that HyperMem achieves state-of-the-art performance with 92.73% LLM-as-a-judge accuracy, demonstrating the effectiveness of HyperMem for long-term conversations.

Figures & tables

Figure 1: Memory structure comparison across Chunk-based RAG, Graph-based RAG, and our HyperMem.
Figure 2: Framework of HyperMem. The indexing detects episode boundaries, aggregates topics via hyperedges, and extracts facts. The retrieval performs coarse-to-fine search from topics to episodes to facts.
MethodsSingle-hopMulti-hopTemporalOpen DomainOverall
GraphRAG Edge et al. (2024)79.5554.9650.1658.3367.60
LightRAG Guo et al. (2025)86.6884.0460.7571.8879.87
HippoRAG 2 Gutiérrez et al. (2025)86.4475.8978.5066.6781.62
HyperGraphRAG Luo et al. (2025)90.6180.8585.3670.8386.49
OpenAI 163.7942.9221.7162.2952.90
LangMem 262.2347.9223.4371.1258.10
Table 1: Comparison of HyperMem with RAG-based and memory system methods on the LoCoMo benchmark. Evaluation metric is LLM-as-a-judge accuracy (%) scored by GPT-4o-mini. †\dagger indicates results obtained using GPT-4.1-mini. Results for RAG-based methods are reproduced using their official implementations. Results for memory systems are primarily sourced from Chhikara et al. (2025) ; Wang and Chen (2025) ; Li et al. (2025b) .
ConfigurationOverall (%)Δ\Delta
HyperMem92.66–
w/o FC91.750.91 ↓\downarrow
w/o EC88.903.76 ↓\downarrow
w/o TR91.940.72 ↓\downarrow
w/o TR & FC91.750.91 ↓\downarrow
w/o TR & EC88.833.83 ↓\downarrow
Table 2: Ablation study. FC: Fact Context, EC: Episode Context, TR: Topic Retrieval, ER: Episode Retrieval.
Figure 3: Ablation study across four question categories. FC: Fact context. EC: Episode context. TR: Topic-level retrieval. ER: Episode-level retrieval. The shaded region highlights the full HyperMem configuration.
Figure 4: Hyperparameter sensitivity analysis on LoCoMo. We evaluate the impact of embedding fusion weight α\alpha and Top-k selection at each hierarchical level (Topic, Episode, Fact) on retrieval performance.
Figure 5: Token usage vs. accuracy comparison. The x-axis shows relative token usage (Mem0 as 1.0 ×\times baseline), and the y-axis shows LLM-as-a-judge accuracy.
Appendix figures & tables8 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 8
Figure 6: Prompt template of episode boundary detection.
Figure 7: Prompt templates of topic aggregation.
Figure 8: Prompt templates of fact extraction.
Figure 9: Single-Hop Task. HyperMem precisely retrieves “dog shelter” while GraphRAG confuses it with “homeless shelter.”
Figure 10: Multi-Hop Task. HyperMem aggregates all 7 tournament mentions across 10 months via Topic hyperedges, while baselines fragment evidence.
Figure 11: Temporal Reasoning Task. HyperMem correctly identifies one pet at the queried time point, while baselines confuse subjects or miscount.
Figure 12: Open Domain Task. HyperMem infers John’s commitment to U.S.-based goals, while HyperGraphRAG incorrectly concludes he would relocate.
More in this topic

Long-Term Conversational Memory

Long-term conversational memory helps AI assistants retain, update, and use information from past conversations across sessions.

See all 267 papers

Explore similar work

CardsList

  1. APEX-MEM: Agentic Semi-Structured Memory with Temporal Reasoning for Long-Term Conversational AI

    Apr 15, 2026Pratyay Banerjee, Masud Moshtaghi, Shivashankar Subramanian +2Long-Term Conversational MemoryGraph-Based Agent Memory

    Amazon, AGI, Sunnyvale, USA

  2. ETAMem: Decoupling and Aligning Heterogeneous Memory for Long-Term Conversational Agents

    Oct 27, 2026Xinying Qian, Ying Zhang, Yu Zhao +3Memory Management for LLM AgentsLong-Term Conversational Memory

    Nankai University, Tianjin, China · Tiangong University, Tianjin, China · The University of Edinburgh, Edinburgh, United Kingdom

  3. HiGMem: A Hierarchical and LLM-Guided Memory System for Long-Term Conversational Agents

    Apr 20, 2026Shuqi Cao, Jingyi He, Fei TanPersistent Memory for Language ModelsHierarchical Memory

    East China Normal University, Shanghai, China · Shanghai Jiao Tong University, Shanghai, China