@inproceedings{wang-etal-2025-contrastscore,
title = "{C}ontrast{S}core: Towards Higher Quality, Less Biased, More Efficient Evaluation Metrics with Contrastive Evaluation",
author = "Wang, Xiao and
Larionov, Daniil and
Wu, Siwei and
Liu, Yiqi and
Eger, Steffen and
Moosavi, Nafise Sadat and
Lin, Chenghua",
editor = "Inui, Kentaro and
Sakti, Sakriani and
Wang, Haofen and
Wong, Derek F. and
Bhattacharyya, Pushpak and
Banerjee, Biplab and
Ekbal, Asif and
Chakraborty, Tanmoy and
Singh, Dhirendra Pratap",
booktitle = "Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics",
month = dec,
year = "2025",
address = "Mumbai, India",
publisher = "The Asian Federation of Natural Language Processing and The Association for Computational Linguistics",
url = "https://aclanthology.org/2025.ijcnlp-long.163/",
pages = "3045--3060",
ISBN = "979-8-89176-298-5",
abstract = "Recent advances in automatic evaluation of natural language generation have increasingly relied on large language models as general-purpose metrics. While effective, these approaches often require high-capacity models, which introduce substantial computational costs, and remain susceptible to known evaluation pathologies, such as over-reliance on likelihood. We introduce ContrastScore, a contrastive evaluation paradigm that builds on the widely used BARTScore formulation by comparing token-level probabilities between a stronger and a weaker model. Instead of relying on single-model likelihoods or prompt-based judgments, ContrastScore captures disagreement between models to better reflect confidence and uncertainty in generation quality. Empirical results on summarization and machine translation benchmarks show that ContrastScore, instantiated with paired moderate-scale models across both Qwen and LLaMA families, consistently outperforms larger alternatives, such as Qwen 7B and LLaMA 8B, in correlation with human ratings. In addition to improving evaluation quality, ContrastScore significantly reduces susceptibility to likelihood bias, offering a more robust and cost-effective alternative to larger LLM-based evaluation methods."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-etal-2025-contrastscore">
<titleInfo>
<title>ContrastScore: Towards Higher Quality, Less Biased, More Efficient Evaluation Metrics with Contrastive Evaluation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xiao</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Daniil</namePart>
<namePart type="family">Larionov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siwei</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yiqi</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Steffen</namePart>
<namePart type="family">Eger</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nafise</namePart>
<namePart type="given">Sadat</namePart>
<namePart type="family">Moosavi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chenghua</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kentaro</namePart>
<namePart type="family">Inui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sakriani</namePart>
<namePart type="family">Sakti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haofen</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Derek</namePart>
<namePart type="given">F</namePart>
<namePart type="family">Wong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pushpak</namePart>
<namePart type="family">Bhattacharyya</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Biplab</namePart>
<namePart type="family">Banerjee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Asif</namePart>
<namePart type="family">Ekbal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dhirendra</namePart>
<namePart type="given">Pratap</namePart>
<namePart type="family">Singh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>The Asian Federation of Natural Language Processing and The Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Mumbai, India</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-298-5</identifier>
</relatedItem>
<abstract>Recent advances in automatic evaluation of natural language generation have increasingly relied on large language models as general-purpose metrics. While effective, these approaches often require high-capacity models, which introduce substantial computational costs, and remain susceptible to known evaluation pathologies, such as over-reliance on likelihood. We introduce ContrastScore, a contrastive evaluation paradigm that builds on the widely used BARTScore formulation by comparing token-level probabilities between a stronger and a weaker model. Instead of relying on single-model likelihoods or prompt-based judgments, ContrastScore captures disagreement between models to better reflect confidence and uncertainty in generation quality. Empirical results on summarization and machine translation benchmarks show that ContrastScore, instantiated with paired moderate-scale models across both Qwen and LLaMA families, consistently outperforms larger alternatives, such as Qwen 7B and LLaMA 8B, in correlation with human ratings. In addition to improving evaluation quality, ContrastScore significantly reduces susceptibility to likelihood bias, offering a more robust and cost-effective alternative to larger LLM-based evaluation methods.</abstract>
<identifier type="citekey">wang-etal-2025-contrastscore</identifier>
<location>
<url>https://aclanthology.org/2025.ijcnlp-long.163/</url>
</location>
<part>
<date>2025-12</date>
<extent unit="page">
<start>3045</start>
<end>3060</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ContrastScore: Towards Higher Quality, Less Biased, More Efficient Evaluation Metrics with Contrastive Evaluation
%A Wang, Xiao
%A Larionov, Daniil
%A Wu, Siwei
%A Liu, Yiqi
%A Eger, Steffen
%A Moosavi, Nafise Sadat
%A Lin, Chenghua
%Y Inui, Kentaro
%Y Sakti, Sakriani
%Y Wang, Haofen
%Y Wong, Derek F.
%Y Bhattacharyya, Pushpak
%Y Banerjee, Biplab
%Y Ekbal, Asif
%Y Chakraborty, Tanmoy
%Y Singh, Dhirendra Pratap
%S Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics
%D 2025
%8 December
%I The Asian Federation of Natural Language Processing and The Association for Computational Linguistics
%C Mumbai, India
%@ 979-8-89176-298-5
%F wang-etal-2025-contrastscore
%X Recent advances in automatic evaluation of natural language generation have increasingly relied on large language models as general-purpose metrics. While effective, these approaches often require high-capacity models, which introduce substantial computational costs, and remain susceptible to known evaluation pathologies, such as over-reliance on likelihood. We introduce ContrastScore, a contrastive evaluation paradigm that builds on the widely used BARTScore formulation by comparing token-level probabilities between a stronger and a weaker model. Instead of relying on single-model likelihoods or prompt-based judgments, ContrastScore captures disagreement between models to better reflect confidence and uncertainty in generation quality. Empirical results on summarization and machine translation benchmarks show that ContrastScore, instantiated with paired moderate-scale models across both Qwen and LLaMA families, consistently outperforms larger alternatives, such as Qwen 7B and LLaMA 8B, in correlation with human ratings. In addition to improving evaluation quality, ContrastScore significantly reduces susceptibility to likelihood bias, offering a more robust and cost-effective alternative to larger LLM-based evaluation methods.
%U https://aclanthology.org/2025.ijcnlp-long.163/
%P 3045-3060
Markdown (Informal)
[ContrastScore: Towards Higher Quality, Less Biased, More Efficient Evaluation Metrics with Contrastive Evaluation](https://aclanthology.org/2025.ijcnlp-long.163/) (Wang et al., IJCNLP-AACL 2025)
ACL
- Xiao Wang, Daniil Larionov, Siwei Wu, Yiqi Liu, Steffen Eger, Nafise Sadat Moosavi, and Chenghua Lin. 2025. ContrastScore: Towards Higher Quality, Less Biased, More Efficient Evaluation Metrics with Contrastive Evaluation. In Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics, pages 3045–3060, Mumbai, India. The Asian Federation of Natural Language Processing and The Association for Computational Linguistics.