@inproceedings{kompally-s-2026-geometry,
title = "Geometry-Conditioned Turn Scoring for Conversational Memory Compression",
author = "Kompally, Pranav and
S, Sibi Chakkaravarthy",
editor = "Wang, Yimu and
Choi, Yee Man and
Wu, Di and
Ouyang, Siqi and
Azadani, Mozhgan Nasr and
Fung, Yi R. (May)",
booktitle = "Proceedings of the 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently ({G}round{LM} 2026)",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.groundlm-1.37/",
pages = "463--471",
ISBN = "979-8-89176-478-1",
abstract = "Compressing long multi-turn conversations requires deciding not only how to compress context, but which signal should rank memory candidates under a fixed budget. We study a sparse keep/compress/drop (KCD) codec and compare geometric, semantic, and hybrid scoring signals across conversational memory regimes. On a stress-test benchmark designed to expose support-turn failures, a geometry-scored codec substantially outperforms a semantic signal swap inside the same codec. On MSC (1,000 conversations), we show that geometry-based compression methods substantially outperform LongLLMLingua on geometric fidelity ({\ensuremath{\Delta}}L2 {\ensuremath{\approx}} +130) while LongLLMLingua modestly improves answer NLL{---}a geometry-behavior divergence confirming that behavioral metrics alone are insufficient to evaluate memory compression. Among KCD signal variants on MSC (1,000 conversations, Qwen2.5-1.5B), query-conditioned geometry achieves the best combination of geometric fidelity and behavioral quality simultaneously: at budget 0.50 it reduces logit L2 by 109.4 units versus semantic KCD (p{\ensuremath{<}}0.001, conversation-level) while also achieving best answer NLL across all budgets. On LongMemEval-S (100 full-length conversations), all KCD variants significantly improve behavioral quality over uniform compression. A Llama-3.2-3B cross-model check shows that the geometry signal advantage is model-family-dependent: semantic compression dominates geometry compression on behavior across all budgets for Llama-3B (p{\ensuremath{\leq}}0.01, conversation-level). The paper{'}s central claim is that geometry-behavior divergence is real and consequential, and that query-conditioned geometry is the signal that best resolves it on Qwen2.5 models."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kompally-s-2026-geometry">
<titleInfo>
<title>Geometry-Conditioned Turn Scoring for Conversational Memory Compression</title>
</titleInfo>
<name type="personal">
<namePart type="given">Pranav</namePart>
<namePart type="family">Kompally</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sibi</namePart>
<namePart type="given">Chakkaravarthy</namePart>
<namePart type="family">S</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently (GroundLM 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yimu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yee</namePart>
<namePart type="given">Man</namePart>
<namePart type="family">Choi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Di</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siqi</namePart>
<namePart type="family">Ouyang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mozhgan</namePart>
<namePart type="given">Nasr</namePart>
<namePart type="family">Azadani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yi</namePart>
<namePart type="given">R</namePart>
<namePart type="given">(May)</namePart>
<namePart type="family">Fung</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-478-1</identifier>
</relatedItem>
<abstract>Compressing long multi-turn conversations requires deciding not only how to compress context, but which signal should rank memory candidates under a fixed budget. We study a sparse keep/compress/drop (KCD) codec and compare geometric, semantic, and hybrid scoring signals across conversational memory regimes. On a stress-test benchmark designed to expose support-turn failures, a geometry-scored codec substantially outperforms a semantic signal swap inside the same codec. On MSC (1,000 conversations), we show that geometry-based compression methods substantially outperform LongLLMLingua on geometric fidelity (\ensuremathΔL2 \ensuremath\approx +130) while LongLLMLingua modestly improves answer NLL—a geometry-behavior divergence confirming that behavioral metrics alone are insufficient to evaluate memory compression. Among KCD signal variants on MSC (1,000 conversations, Qwen2.5-1.5B), query-conditioned geometry achieves the best combination of geometric fidelity and behavioral quality simultaneously: at budget 0.50 it reduces logit L2 by 109.4 units versus semantic KCD (p\ensuremath<0.001, conversation-level) while also achieving best answer NLL across all budgets. On LongMemEval-S (100 full-length conversations), all KCD variants significantly improve behavioral quality over uniform compression. A Llama-3.2-3B cross-model check shows that the geometry signal advantage is model-family-dependent: semantic compression dominates geometry compression on behavior across all budgets for Llama-3B (p\ensuremathłeq0.01, conversation-level). The paper’s central claim is that geometry-behavior divergence is real and consequential, and that query-conditioned geometry is the signal that best resolves it on Qwen2.5 models.</abstract>
<identifier type="citekey">kompally-s-2026-geometry</identifier>
<location>
<url>https://aclanthology.org/2026.groundlm-1.37/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>463</start>
<end>471</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Geometry-Conditioned Turn Scoring for Conversational Memory Compression
%A Kompally, Pranav
%A S, Sibi Chakkaravarthy
%Y Wang, Yimu
%Y Choi, Yee Man
%Y Wu, Di
%Y Ouyang, Siqi
%Y Azadani, Mozhgan Nasr
%Y Fung, Yi R. (May)
%S Proceedings of the 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently (GroundLM 2026)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-478-1
%F kompally-s-2026-geometry
%X Compressing long multi-turn conversations requires deciding not only how to compress context, but which signal should rank memory candidates under a fixed budget. We study a sparse keep/compress/drop (KCD) codec and compare geometric, semantic, and hybrid scoring signals across conversational memory regimes. On a stress-test benchmark designed to expose support-turn failures, a geometry-scored codec substantially outperforms a semantic signal swap inside the same codec. On MSC (1,000 conversations), we show that geometry-based compression methods substantially outperform LongLLMLingua on geometric fidelity (\ensuremathΔL2 \ensuremath\approx +130) while LongLLMLingua modestly improves answer NLL—a geometry-behavior divergence confirming that behavioral metrics alone are insufficient to evaluate memory compression. Among KCD signal variants on MSC (1,000 conversations, Qwen2.5-1.5B), query-conditioned geometry achieves the best combination of geometric fidelity and behavioral quality simultaneously: at budget 0.50 it reduces logit L2 by 109.4 units versus semantic KCD (p\ensuremath<0.001, conversation-level) while also achieving best answer NLL across all budgets. On LongMemEval-S (100 full-length conversations), all KCD variants significantly improve behavioral quality over uniform compression. A Llama-3.2-3B cross-model check shows that the geometry signal advantage is model-family-dependent: semantic compression dominates geometry compression on behavior across all budgets for Llama-3B (p\ensuremathłeq0.01, conversation-level). The paper’s central claim is that geometry-behavior divergence is real and consequential, and that query-conditioned geometry is the signal that best resolves it on Qwen2.5 models.
%U https://aclanthology.org/2026.groundlm-1.37/
%P 463-471
Markdown (Informal)
[Geometry-Conditioned Turn Scoring for Conversational Memory Compression](https://aclanthology.org/2026.groundlm-1.37/) (Kompally & S, GroundLM 2026)
ACL