Don't Scroll Back: Missing-Evidence Memory for Streaming Dialogue Summarization
Summary
This paper introduces ReMEMBER, a missing-evidence memory framework for streaming dialogue summarization that retrieves and refines evidence from long histories to resolve gaps in current windows under fixed memory budgets, along with a benchmark for evaluation.
View Cached Full Text
Cached at: 08/12/26, 08:23 AM
Paper page - Don’t Scroll Back: Missing-Evidence Memory for Streaming Dialogue Summarization
Source: https://huggingface.co/papers/2608.09043
Abstract
A framework for streaming dialogue summarization retrieves and refines evidence from long histories to resolve missing context in current windows under fixed memory budgets.
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 asstreaming dialogue summarization, where a system must summarize a current window usingselective memoryfrom 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 containsgap-resolving evidenceand whether the generated summary reflects it. We proposeReMEMBER, amissing-evidence memory frameworkthat conditions retrieval on unresolved window dependencies and refines retrieved chunks intoevidence-dense memoryunder a fixed budget. Experiments on dialogues with histories up to 160K tokens show thatReMEMBERimproves memory recall and gap-resolution completeness over memory construction baselines under the same budget.
View arXiv pageView PDFAdd to collection
Community
Paper author
Paper submitter
Upload images, audio, and videos by dragging in the text input, pasting, orclicking here.
Tap or paste here to upload images
Get this paper in your agent:
hf papers read 2608\.09043
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2608.09043 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2608.09043 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2608.09043 in a Space README.md to link it from this page.
Collections including this paper1
Similar Articles
Detect, Remask, Repair: Diffusion Editing for Faithful Summarization of Evolving Contexts
This paper proposes Detect–Remask–Repair, a diffusion-based framework for localized faithfulness repair in summarization when contexts evolve, and introduces the StreamSum benchmark for evaluating such settings. Experiments show it offers controllable trade-offs between faithfulness, speed, and content preservation.
Reasoning with Memory: A Temporal Granularity-Adaptive Framework for Training-Free Long Video Understanding
ReMem introduces a dual-level memory-augmented keyframe selection framework for training-free long video understanding, achieving state-of-the-art zero-shot performance on multiple benchmarks.
EviMem: Evidence-Gap-Driven Iterative Retrieval for Long-Term Conversational Memory
EviMem combines IRIS for evidence-gap detection and LaceMem for layered memory to improve long-term conversational memory retrieval, achieving higher accuracy on temporal and multi-hop questions with lower latency.
StreamMemBench: Streaming Evaluation of Agent Memory for Future-Oriented Assistance
StreamMemBench is a new streaming benchmark that tests how well personal-agent memory systems use observed evidence and user feedback for future-oriented assistance. Experiments show current systems often fail to turn stored information into reliable follow-up behavior.
Echo-Forcing: A Scene Memory Framework for Interactive Long Video Generation
Echo-Forcing introduces a scene memory framework for interactive long video generation, using hierarchical temporal memory, scene recall frames, and difference-aware memory decay to handle prompt switching and long-term recall. The method is training-free and achieves strong performance on VBench-Long.