@inproceedings{fan-etal-2026-patalign,
title = "{P}at{A}lign: Bridging the Gap between Scientific Papers and Patent Descriptions via Iterative Preference Optimization",
author = "Fan, Zheming and
Song, Xuemeng and
Yang, Min and
Yang, Linyi",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.212/",
pages = "4434--4448",
ISBN = "979-8-89176-462-0",
abstract = "Transforming academic papers into patent descriptions challenges LLMs with extreme length, strict factuality, and rigorous stylistic norms. Existing supervised methods suffer from pathological repetition, while static preference optimization causes style degradation by erroneously penalizing necessary structural repetitions. To address this, we propose \textsc{PatAlign}, a framework utilizing Iterative Preference Optimization to dynamically sample negative instances from the evolving policy, effectively distinguishing pathological repetition from the structural repetition required to ensure patent rigor. We also introduce the Paper-to-Patent Score (P$^2$S) metric to evaluate the overall quality of generated patents. Experiments demonstrate that \textsc{PatAlign} effectively mitigates pathological repetition and style degradation, achieving state-of-the-art performance in both factual consistency and language style. Notably, our 8B model significantly outperforms 32B general-purpose models, validating the efficacy of iterative alignment strategies in complex long-form generation tasks. Our code and data are available at \url{https://github.com/SUSTech-GenAI/PatAlign}."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="fan-etal-2026-patalign">
<titleInfo>
<title>PatAlign: Bridging the Gap between Scientific Papers and Patent Descriptions via Iterative Preference Optimization</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zheming</namePart>
<namePart type="family">Fan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xuemeng</namePart>
<namePart type="family">Song</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Min</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Linyi</namePart>
<namePart type="family">Yang</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>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Transforming academic papers into patent descriptions challenges LLMs with extreme length, strict factuality, and rigorous stylistic norms. Existing supervised methods suffer from pathological repetition, while static preference optimization causes style degradation by erroneously penalizing necessary structural repetitions. To address this, we propose PatAlign, a framework utilizing Iterative Preference Optimization to dynamically sample negative instances from the evolving policy, effectively distinguishing pathological repetition from the structural repetition required to ensure patent rigor. We also introduce the Paper-to-Patent Score (P²S) metric to evaluate the overall quality of generated patents. Experiments demonstrate that PatAlign effectively mitigates pathological repetition and style degradation, achieving state-of-the-art performance in both factual consistency and language style. Notably, our 8B model significantly outperforms 32B general-purpose models, validating the efficacy of iterative alignment strategies in complex long-form generation tasks. Our code and data are available at https://github.com/SUSTech-GenAI/PatAlign.</abstract>
<identifier type="citekey">fan-etal-2026-patalign</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.212/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>4434</start>
<end>4448</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T PatAlign: Bridging the Gap between Scientific Papers and Patent Descriptions via Iterative Preference Optimization
%A Fan, Zheming
%A Song, Xuemeng
%A Yang, Min
%A Yang, Linyi
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F fan-etal-2026-patalign
%X Transforming academic papers into patent descriptions challenges LLMs with extreme length, strict factuality, and rigorous stylistic norms. Existing supervised methods suffer from pathological repetition, while static preference optimization causes style degradation by erroneously penalizing necessary structural repetitions. To address this, we propose PatAlign, a framework utilizing Iterative Preference Optimization to dynamically sample negative instances from the evolving policy, effectively distinguishing pathological repetition from the structural repetition required to ensure patent rigor. We also introduce the Paper-to-Patent Score (P²S) metric to evaluate the overall quality of generated patents. Experiments demonstrate that PatAlign effectively mitigates pathological repetition and style degradation, achieving state-of-the-art performance in both factual consistency and language style. Notably, our 8B model significantly outperforms 32B general-purpose models, validating the efficacy of iterative alignment strategies in complex long-form generation tasks. Our code and data are available at https://github.com/SUSTech-GenAI/PatAlign.
%U https://aclanthology.org/2026.findings-emnlp.212/
%P 4434-4448
Markdown (Informal)
[PatAlign: Bridging the Gap between Scientific Papers and Patent Descriptions via Iterative Preference Optimization](https://aclanthology.org/2026.findings-emnlp.212/) (Fan et al., Findings 2026)
ACL