Don't Scroll Back: Missing-Evidence Memory for Streaming Dialogue Summarization

2026-08-09

AI Summary

This work addresses the challenge of insufficient contextual information within the current window in streaming dialogue summarization, which necessitates efficient extraction of critical evidence from extensive dialogue histories. The authors propose a memory mechanism specifically designed to bridge semantic gaps: it identifies missing information in the current window through dependency analysis and selectively retrieves and compresses relevant historical content to construct an evidence-dense memory under a fixed budget. Departing from naive strategies that retain arbitrary historical segments, the approach focuses on precisely supplementing semantically dependent elements. A new benchmark is also introduced to evaluate memory-based evidence coverage and summary faithfulness. Experiments demonstrate that, over dialogue histories up to 160K tokens, the proposed method significantly outperforms baseline approaches, achieving higher memory recall and more complete coverage of gap-related evidence within the same memory budget.

Abstract

Users of modern platforms repeatedly need summaries of recent dialogue, but the window rarely contains enough context to be interpreted on its own. We formalize this setting as streaming dialogue summarization, where a system must summarize a current window using selective memory from an unbounded history under a fixed budget. We show that the central challenge is not how much history is accessed, but whether memory recovers the evidence that the current window presupposes. We construct a benchmark and evaluation protocol that separately assesses whether memory contains gap-resolving evidence and whether the generated summary reflects it. We propose ReMEMBER, a missing-evidence memory framework that conditions retrieval on unresolved window dependencies and refines retrieved chunks into evidence-dense memory under a fixed budget. Experiments on dialogues with histories up to 160K tokens show that ReMEMBER improves memory recall and gap-resolution completeness over memory construction baselines under the same budget.

Problem

• streaming dialogue summarization
• missing-evidence memory
• context gap
• memory budget
• evidence retrieval

Innovation

• streaming dialogue summarization
• missing-evidence memory
• evidence retrieval
• memory budgeting
• gap-resolution