Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ActMem: Bridging the Gap Between Memory Retrieval and Reasoning in LLM Agents

Xiaohui Zhang, Zequn Sun, Chengyuan Yang, Yaqin Jin, Yazhong Zhang, Wei Hu

TL;DR

A novel actionable memory framework called ActMem is proposed that integrates memory retrieval with active causal reasoning and enables agents to deduce implicit constraints and resolve potential conflicts between past states and current intentions.

Abstract

Effective memory management is essential for large language model (LLM) agents handling long-term interactions. Current memory frameworks typically treat agents as passive "recorders" and retrieve information without understanding its deeper implications. They may fail in scenarios requiring conflict detection and complex decision-making. To bridge this critical gap, we propose a novel actionable memory framework called ActMem that integrates memory retrieval with active causal reasoning. ActMem transforms unstructured dialogue history into a structured causal and semantic graph. By leveraging counterfactual reasoning and commonsense completion, it enables agents to deduce implicit constraints and resolve potential conflicts between past states and current intentions. Furthermore, we introduce a comprehensive dataset ActMemEval to evaluate agent reasoning capabilities in logic-driven scenarios, moving beyond the fact-retrieval focus of existing memory benchmarks. Experiments demonstrate that ActMem significantly outperforms state-of-the-art baselines in handling complex, memory-dependent tasks, paving the way for more consistent and reliable intelligent assistants.

ActMem: Bridging the Gap Between Memory Retrieval and Reasoning in LLM Agents

TL;DR

A novel actionable memory framework called ActMem is proposed that integrates memory retrieval with active causal reasoning and enables agents to deduce implicit constraints and resolve potential conflicts between past states and current intentions.

Abstract

Effective memory management is essential for large language model (LLM) agents handling long-term interactions. Current memory frameworks typically treat agents as passive "recorders" and retrieve information without understanding its deeper implications. They may fail in scenarios requiring conflict detection and complex decision-making. To bridge this critical gap, we propose a novel actionable memory framework called ActMem that integrates memory retrieval with active causal reasoning. ActMem transforms unstructured dialogue history into a structured causal and semantic graph. By leveraging counterfactual reasoning and commonsense completion, it enables agents to deduce implicit constraints and resolve potential conflicts between past states and current intentions. Furthermore, we introduce a comprehensive dataset ActMemEval to evaluate agent reasoning capabilities in logic-driven scenarios, moving beyond the fact-retrieval focus of existing memory benchmarks. Experiments demonstrate that ActMem significantly outperforms state-of-the-art baselines in handling complex, memory-dependent tasks, paving the way for more consistent and reliable intelligent assistants.
Paper Structure (23 sections, 7 equations, 7 figures, 3 tables)

This paper contains 23 sections, 7 equations, 7 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Agent Memory Management
  4. Synthetic Evaluation Benchmark
  5. The ActMem Framework
  6. Memory Fact Extraction
  7. Fact Clustering
  8. Memory KG Construction
  9. Counterfactual-based Retrieval and Reasoning
  10. The ActMemEval Benchmark
  11. Dataset Construction
  12. Dataset Evaluation
  13. Experiments
  14. Baselines
  15. Evaluation Metrics
  16. ...and 8 more sections

Figures (7)

  • Figure 1: A motivating comparison between memory retrieval and our proposed memory reasoning setting. (a) An example from LongMemEval representing memory retrieval. (b) An example illustrating the necessity of memory reasoning. Although the user's current query (buying Sago Palms) does not semantically overlap with the past memory (a teething puppy), the agent should infer the latent conflict based on the common sense that Sago Palms are toxic.
  • Figure 2: The overview of our proposed framework.
  • Figure 3: Pipeline to construct our ActMemEval dataset.
  • Figure 4: Distribution of semantic similarity scores between the ground-truth answers and their corresponding context history.
  • Figure 5: Ablation study on counterfactual reasoning, causal edge construction and semantic edge construction.
  • ...and 2 more figures