FTA-Mem: Fact-Time-Affect Anchored Memory for Low-Density Long-Term Dialogue

2026-08-17

AI Summary

This study addresses memory challenges in long-term emotional support conversations arising from information sparsity, scattered evidence, and evolving user states by proposing the FTA-Mem framework. The method employs boundary-preserving window segmentation to construct situational segments and designs FTA memory units integrating factual, temporal, and affective dimensions, enabling precise retrieval and comprehension through structured context synthesis. This approach effectively balances evidence retention with computational efficiency. Evaluations on the ES-MemEval and LoCoMo benchmarks demonstrate significant improvements in question-answering performance, achieving an F1 score of 0.3871 and a BERTScore of 0.6668. These results validate the framework’s effectiveness in modeling long-term personalized memory for emotionally supportive dialogue systems.

Abstract

Long-term emotional-support agents require memory mechanisms for personalized understanding across sessions. However, emotional-support dialogue is often low-density: turns are incomplete, evidence is scattered, and user states evolve over time. Existing memory methods usually rely on fixed units, such as turn-level notes or session summaries, which may lose details or introduce redundant noise. We propose FTA-Mem, a structured memory framework for low-density long-term dialogue. FTA-Mem uses Boundary-preserving Window Segmentation (BWS) to form coherent situation fragments, and constructs Fact-Time-Affect Memory Units (FTA Units) that jointly encode factual content, temporal grounding, and affective context. Retrieved units are then synthesized into structured context for answer generation. Experiments on ES-MemEval and LoCoMo show that FTA-Mem improves overall long-term memory question answering across benchmarks with different information-density characteristics. On ES-MemEval, FTA-Mem achieves 0.3871 F1 and 0.6668 BERTScore. Further analysis shows that situation-level FTA construction better balances evidence preservation and construction cost than coarse session-level or overly fine-grained turn-pair construction, providing an effective granularity trade-off for long-term dialogue memory.

Problem

• Long-term dialogue
• Emotional support
• Low-density
• Memory mechanism
• Personalized understanding

Innovation

• Fact-Time-Affect Memory Units
• Boundary-preserving Window Segmentation
• Low-Density Long-Term Dialogue
• Structured Memory Framework
• Situation-level Granularity