cs.AIMar 29, 2026

GAAMA: Graph Augmented Associative Memory for Agents

Authors: Swarna Kamal Paul, Shubhendu Sharma, Nitin Sareen

Organizations: Nagarro

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

AI agents that interact with users across multiple sessions require persistent long-term memory to maintain coherent, personalized behavior. Current approaches either rely on flat retrieval-augmented generation (RAG), which loses structural relationships among memories, or use entity-centric knowledge graphs that suffer from mega-hub effects in conversational data, diluting graph-based relevance propagation. We propose GAAMA, a graph-augmented associative memory for agents that constructs a concept-mediated knowledge graph through a three-step pipeline: (1)verbatim episode preservation, (2)LLM-based extraction of atomic facts and topic-level concept nodes, and (3)synthesis of higher-order reflections. The resulting graph uses four node types (episode, fact, reflection, concept) connected by five structural edge types, with concept nodes providing cross-cutting traversal paths that avoid the mega-hub problem of entity-centric designs. Retrieval combines cosine-similarity-based k-nearest neighbor search with edge-type-aware Personalized PageRank (PPR) through an additive scoring function. We further introduce GRAFT (Graph Repair by Augmenting Facts & Topology), a post-retrieval corrective layer that diagnoses retrieval failures and surgically repairs the knowledge graph. On LoCoMo-10 (1,540 questions, 10 multi-session conversations), GAAMA achieves 79.1% mean reward, a +4.2pp improvement over a tuned RAG baseline, the strongest comparator. On MemoryArena, GAAMA outperforms full-context baselines across three tasks - Group Travel (+0.4pp), Web Shopping (+3.4pp), and Progressive Search (+0.7pp) - with advantages growing monotonically with dialogue length. Notably, GAAMA delivers consistent performance across all categories, matching the best competing method in each, whereas every competitor degrades in at least one category.

Figures & tables

Edge TypeSource →\to TargetSemantics
NEXTEpisode →\to EpisodeTemporal succession within a session
DERIVED_FROMFact →\to EpisodeProvenance: fact extracted from episode
DERIVED_FROM_FACTReflection →\to FactProvenance: reflection from facts
HAS_CONCEPTEpisode →\to ConceptEpisode discusses topic
ABOUT_CONCEPTFact →\to ConceptFact relates to topic
Table 1: Graph schema: four node types connected by five structural edge types.
Edge TypeWeightEdge TypeWeight
NEXT0.8HAS_CONCEPT0.8
DERIVED_FROM0.8ABOUT_CONCEPT0.8
DERIVED_FROM_FACT0.5  
Table 2: Edge-type base weights for PPR transition probability computation.
SystemMulti-hop (Cat1)Temporal (Cat2)Open Domain (Cat3)Single Hop (Cat4)Overall
A-Mem44.737.450.051.547.2
Nemori49.445.036.857.452.1
Mem051.255.572.967.162.1
HippoRAG61.767.067.774.169.9
RAG Baseline67.559.044.686.974.9
GAAMA (ours)72.475.147.386.679.1
Table 3: Mean reward (%) on LoCoMo-10 by question category. Best in bold . Approximate overall computed using our category weights.
CategorynnEditsNoiseGRAFT ++GRAFT −-Net GRAFT
Cat1 (Multi-hop)28221−- 0.2++ 4.6−- 2.1++ 2.5
Cat2 (Temporal)32116−- 0.4++ 2.1−- 4.8−- 2.7
Cat3 (Open Domain)9618−- 0.2++ 6.4−- 5.4++ 1.0
Cat4 (Single Hop)84126++ 0.0++ 2.1−- 1.8++ 0.3
Overall154081−- 0.1++ 2.8−- 2.7++ 0.1
Table 4: GRAFT reward attribution per category (pp). Noise estimated from zero-edit conversations.
Conv.Cat1Cat2Cat3Cat4Overall
conv-26−- 4.9++ 4.5++ 5.2++ 5.4++ 3.0
conv-30++ 2.2−- 2.6—++ 1.9++ 0.5
conv-41−- 3.2++ 13.0−- 9.4−- 1.2++ 0.5
conv-42−- 2.3++ 7.1−- 9.1++ 0.6++ 0.8
conv-43−- 1.1−- 1.9−- 1.2++ 0.5−- 0.3
conv-44−- 0.6++ 7.0++ 7.1++ 2.5++ 2.9
Table 5: Heatmap of reward delta (pp): PPR=0.1 −- Semantic, per conversation and category. >+5>+5 +2+2 to +5+5 00 to +2+2 −2-2 to 00 −5-5 to −2-2 <−5<-5
 Travel PlannerWeb ShoppingProg. Search
SystemQ1–Q3Q4–Q7Q1–Q2Q3–Q5Q1–Q4Q5–Q8
HippoRAG69.469.664.162.12.42.2
Mem075.676.772.168.267.872.6
Nemori67.968.760.761.30.70.6
A-Mem71.975.563.945.273.282.0
Full-Context69.971.568.169.871.781.3
Table 6: MemoryArena: average reward (%) for first and second half of steps across all three categories. Q0 excluded. Travel Planner: Q1–Q3 vs Q4–Q7; Web Shopping: Q1–Q2 vs Q3–Q5; Progressive Search: Q1–Q4 vs Q5–Q8.
More in this topic

Graph-Based Agent Memory

Graph-based agent memory organizes an agent’s experiences, facts, and procedures as linked structures for persistent, evidence-aware retrieval and reuse.

See all 133 papers

Explore similar work

CardsList
Aug 29, 2026

Selective Forgetting: A Graph-Based Memory Framework for Long-Term LLM Agents

Knowledge graphs have been proposed as a structured alternative to flat retrieval-augmented generation for long-term agent memory, on the assumption that representing conversations as entities and relations improves recall. We evaluate that assumption directly. Our framework extracts each conversational turn into typed nodes and attributed edges, answers questions from a two-hop subgraph, and periodically prunes nodes that score low on a weighted combination of recency, access frequency, degree centrality, and age. On LongMemEval, the graph does not outperform a flat vector baseline at a matched candidate-generation budget of five retrieval roots: token F1 is 0.4170.417 against 0.4680.468, and a paired bootstrap over 500 questions gives Δ=−0.050Δ= -0.050 (95% CI [−0.085,−0.016][-0.085, -0.016]). The gap is widest on questions that require recalling a specific prior assistant turn, where judged correctness falls from 0.9110.911 to 0.6070.607, suggesting that decomposing a turn into entities discards the surface form these questions depend on. The forgetting module is more successful. Applied once to a persistent 27{,}021-node graph, it removes 9.8% of nodes and 9.5% of stored bytes; token F1 is unchanged (+0.001+0.001, 95% CI [−0.015,+0.016][-0.015, +0.016]) and judged correctness falls by 1.61.6 points, with the 95% interval bounding any loss at 3.83.8 points ([−0.038,+0.006][-0.038, +0.006]). Because our extractor is a single small model evaluated on one benchmark, these results characterise this extraction-based pipeline rather than graph-structured memory in general. Code: https://github.com/skhanzad/Selective-Amnesia

Theo Rusu, Sourena Khanzadeh, Manar Alalfi

Department of Computer Science Toronto Metropolitan University Toronto, Ontario, Canada · The Creative School Toronto Metropolitan University Toronto, Ontario, Canada · Flybits, Creative AI Hub Toronto, Ontario, Canada

Graph-Based Agent MemoryLong-Term Conversational MemoryAgent Memory Management
AskPDF
Apr 14, 2026

GAM: Hierarchical Graph-based Agentic Memory for LLM Agents

To sustain coherent long-term interactions, Large Language Model (LLM) agents must navigate the tension between acquiring new information and retaining prior knowledge. Current unified stream-based memory systems facilitate context updates but remain vulnerable to interference from transient noise. Conversely, discrete structured memory architectures provide robust knowledge retention but often struggle to adapt to evolving narratives. To address this, we propose GAM, a hierarchical Graph-based Agentic Memory framework that explicitly decouples memory encoding from consolidation to effectively resolve the conflict between rapid context perception and stable knowledge retention. By isolating ongoing dialogue in an event progression graph and integrating it into a topic associative network only upon semantic shifts, our approach minimizes interference while preserving long-term consistency. Additionally, we introduce a graph-guided, multi-factor retrieval strategy to enhance context precision. Experiments on LoCoMo and LongDialQA indicate that our method consistently outperforms state-of-the-art baselines in both reasoning accuracy and efficiency.

Zhaofen Wu, Hanrong Zhang, Fulin Lin +9

Zhejiang University · University of Illinois Chicago · Rutgers University +2

Agent Memory ManagementGraph-Based Agent MemoryLong-Term Conversational MemoryMemory-Augmented Language Model Agents
AskPDF
Sep 29, 2026

TAGGRAPH: Tag-Augmented Graphs for Graph Retrieval of Agent Persistent Histories

Long-term memory lets LLM agents recall past interactions and remain consistent across sessions, but memory systems are hard to compare because they often vary in representation, indexing, retrieval, and evaluation. We present a controlled evaluation framework based on shared 5W-style conversational memories. Localized graph configurations traverse a common base graph; AdaptiveGraph adds chronological edges and Personalized PageRank diffusion. We also evaluate BM25 over the same extracted notes and OpenClaw as a raw-input external reference. Retrieval rankings vary across memory settings. On LongMemEval-S, AdaptiveGraph is the strongest graph configuration at 0.844 MRR, but BM25 reaches 0.867 and OpenClaw 0.880. On ATANT Core, localized graph traversal outperforms diffusion and BM25, whereas BM25 leads the stress rounds. Reducing LongMemEval-S within the tested range does not reproduce the ATANT diffusion penalty, but the smallest tested store remains larger than ATANT Core, so store size cannot be ruled out. The penalty also persists under a permissive content-match criterion. Vocabulary normalization and extraction quality substantially affect graph retrieval, and missing extraction tags are common among top-five misses. Retrieval strategies should therefore be evaluated jointly with the memory setting and against strong lexical baselines.

Yu-Shu Chen, Yu-Jung Liang, Pengtao Xie

Department of Electrical and Computer Engineering University of California, San Diego La Jolla, CA 92093, USA

Graph-Based RetrievalMemory Agent BenchmarksGraph-Based Agent MemoryLong-Term Conversational Memory
AskPDF