Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Lifelong Dialogue Agents via Timeline-based Memory Management

Kai Tzu-iunn Ong, Namyoung Kim, Minju Gwak, Hyungjoo Chae, Taeyoon Kwon, Yohan Jo, Seung-won Hwang, Dongha Lee, Jinyoung Yeo

TL;DR

The paper addresses the challenge of lifelong dialogue by proposing THEANINE, a timeline-based memory management framework that preserves and links memories through a relation-aware memory graph and retrieves complete memory timelines to augment response generation. It introduces Phase I (memory graph construction), Phase II (timeline retrieval and refinement), and Phase III (timeline-augmented RG), plus TeaFarm, a counterfactual evaluation pipeline for memory recall without ground-truth mappings. Across MSC and CC datasets, Theanine outperforms baselines in automatic and human evaluations, with ablations confirming the importance of relation-aware linking, timeline retrieval, and refinement. The work demonstrates improved retrieval quality and past-entailing responses, offering a scalable, cost-aware path toward robust, personalized, lifelong dialogue agents and a practical evaluation framework for memory integration.

Abstract

To achieve lifelong human-agent interaction, dialogue agents need to constantly memorize perceived information and properly retrieve it for response generation (RG). While prior studies focus on getting rid of outdated memories to improve retrieval quality, we argue that such memories provide rich, important contextual cues for RG (e.g., changes in user behaviors) in long-term conversations. We present THEANINE, a framework for LLM-based lifelong dialogue agents. THEANINE discards memory removal and manages large-scale memories by linking them based on their temporal and cause-effect relation. Enabled by this linking structure, THEANINE augments RG with memory timelines - series of memories representing the evolution or causality of relevant past events. Along with THEANINE, we introduce TeaFarm, a counterfactual-driven evaluation scheme, addressing the limitation of G-Eval and human efforts when assessing agent performance in integrating past memories into RG. A supplementary video for THEANINE and data for TeaFarm are at https://huggingface.co/spaces/ResearcherScholar/Theanine.

Towards Lifelong Dialogue Agents via Timeline-based Memory Management

TL;DR

The paper addresses the challenge of lifelong dialogue by proposing THEANINE, a timeline-based memory management framework that preserves and links memories through a relation-aware memory graph and retrieves complete memory timelines to augment response generation. It introduces Phase I (memory graph construction), Phase II (timeline retrieval and refinement), and Phase III (timeline-augmented RG), plus TeaFarm, a counterfactual evaluation pipeline for memory recall without ground-truth mappings. Across MSC and CC datasets, Theanine outperforms baselines in automatic and human evaluations, with ablations confirming the importance of relation-aware linking, timeline retrieval, and refinement. The work demonstrates improved retrieval quality and past-entailing responses, offering a scalable, cost-aware path toward robust, personalized, lifelong dialogue agents and a practical evaluation framework for memory integration.

Abstract

To achieve lifelong human-agent interaction, dialogue agents need to constantly memorize perceived information and properly retrieve it for response generation (RG). While prior studies focus on getting rid of outdated memories to improve retrieval quality, we argue that such memories provide rich, important contextual cues for RG (e.g., changes in user behaviors) in long-term conversations. We present THEANINE, a framework for LLM-based lifelong dialogue agents. THEANINE discards memory removal and manages large-scale memories by linking them based on their temporal and cause-effect relation. Enabled by this linking structure, THEANINE augments RG with memory timelines - series of memories representing the evolution or causality of relevant past events. Along with THEANINE, we introduce TeaFarm, a counterfactual-driven evaluation scheme, addressing the limitation of G-Eval and human efforts when assessing agent performance in integrating past memories into RG. A supplementary video for THEANINE and data for TeaFarm are at https://huggingface.co/spaces/ResearcherScholar/Theanine.
Paper Structure (52 sections, 8 equations, 29 figures, 9 tables)

This paper contains 52 sections, 8 equations, 29 figures, 9 tables.

Table of Contents

  1. Introduction
  2. Methodologies
  3. Memory Graph Construction (Phase I)
  4. Phase I-1: Identifying associative memories for memory linking.
  5. Phase I-2: Relation-aware memory linking.
  6. Timeline Retrieval and Timeline Refinement (Phase II)
  7. Preparation: Top-k memory retrieval.
  8. Phase II-1: Retrieving and untangling raw memory timelines.
  9. Phase II-2: Context-aware timeline refinement.
  10. Timeline-augmented Response Generation (Phase III)
  11. Experimental Setups
  12. Datasets of Long-term Conversations
  13. Baselines
  14. Models and Implementation Details
  15. Evaluation Scheme 1: Automatic and Human Evaluations
  16. ...and 37 more sections

Figures (29)

  • Figure 1: Empirical examples of failed responses due to (a) absence of an important past event ("afraid of cruise ships") on the timeline and (b) bias to the latest input. (c) is a response augmented with the memory timeline.
  • Figure 2: The overview of Theanine. Left: Linking new memories to the memory graph after finishing a dialogue session; Right: Memory timeline retrieval, refinement, and response generation in a new dialogue session.
  • Figure 3: Locating memories to be linked to $m_{new}$.
  • Figure 4: Extracting raw memory timelines $\tau$ from the retrieved connected component $C_{re}$.
  • Figure 5: Human- (right) and machine-based (left) head-to-head comparisons between ours and baselines regarding the helpfulness of retrieved memories.
  • ...and 24 more figures