@inproceedings{treviso-etal-2021-ist,
title = "{IST}-Unbabel 2021 Submission for the Explainable Quality Estimation Shared Task",
author = "Treviso, Marcos and
Guerreiro, Nuno M. and
Rei, Ricardo and
Martins, Andr{\'e} F. T.",
editor = "Gao, Yang and
Eger, Steffen and
Zhao, Wei and
Lertvittayakumjorn, Piyawat and
Fomicheva, Marina",
booktitle = "Proceedings of the 2nd Workshop on Evaluation and Comparison of NLP Systems",
month = nov,
year = "2021",
address = "Punta Cana, Dominican Republic",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.eval4nlp-1.14",
doi = "10.18653/v1/2021.eval4nlp-1.14",
pages = "133--145",
abstract = "We present the joint contribution of Instituto Superior T{\'e}cnico (IST) and Unbabel to the Explainable Quality Estimation (QE) shared task, where systems were submitted to two tracks: constrained (without word-level supervision) and unconstrained (with word-level supervision). For the constrained track, we experimented with several explainability methods to extract the relevance of input tokens from sentence-level QE models built on top of multilingual pre-trained transformers. Among the different tested methods, composing explanations in the form of attention weights scaled by the norm of value vectors yielded the best results. When word-level labels are used during training, our best results were obtained by using word-level predicted probabilities. We further improve the performance of our methods on the two tracks by ensembling explanation scores extracted from models trained with different pre-trained transformers, achieving strong results for in-domain and zero-shot language pairs.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="treviso-etal-2021-ist">
<titleInfo>
<title>IST-Unbabel 2021 Submission for the Explainable Quality Estimation Shared Task</title>
</titleInfo>
<name type="personal">
<namePart type="given">Marcos</namePart>
<namePart type="family">Treviso</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nuno</namePart>
<namePart type="given">M</namePart>
<namePart type="family">Guerreiro</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ricardo</namePart>
<namePart type="family">Rei</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">André</namePart>
<namePart type="given">F</namePart>
<namePart type="given">T</namePart>
<namePart type="family">Martins</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2021-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2nd Workshop on Evaluation and Comparison of NLP Systems</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yang</namePart>
<namePart type="family">Gao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Steffen</namePart>
<namePart type="family">Eger</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Piyawat</namePart>
<namePart type="family">Lertvittayakumjorn</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marina</namePart>
<namePart type="family">Fomicheva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Punta Cana, Dominican Republic</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>We present the joint contribution of Instituto Superior Técnico (IST) and Unbabel to the Explainable Quality Estimation (QE) shared task, where systems were submitted to two tracks: constrained (without word-level supervision) and unconstrained (with word-level supervision). For the constrained track, we experimented with several explainability methods to extract the relevance of input tokens from sentence-level QE models built on top of multilingual pre-trained transformers. Among the different tested methods, composing explanations in the form of attention weights scaled by the norm of value vectors yielded the best results. When word-level labels are used during training, our best results were obtained by using word-level predicted probabilities. We further improve the performance of our methods on the two tracks by ensembling explanation scores extracted from models trained with different pre-trained transformers, achieving strong results for in-domain and zero-shot language pairs.</abstract>
<identifier type="citekey">treviso-etal-2021-ist</identifier>
<identifier type="doi">10.18653/v1/2021.eval4nlp-1.14</identifier>
<location>
<url>https://aclanthology.org/2021.eval4nlp-1.14</url>
</location>
<part>
<date>2021-11</date>
<extent unit="page">
<start>133</start>
<end>145</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T IST-Unbabel 2021 Submission for the Explainable Quality Estimation Shared Task
%A Treviso, Marcos
%A Guerreiro, Nuno M.
%A Rei, Ricardo
%A Martins, André F. T.
%Y Gao, Yang
%Y Eger, Steffen
%Y Zhao, Wei
%Y Lertvittayakumjorn, Piyawat
%Y Fomicheva, Marina
%S Proceedings of the 2nd Workshop on Evaluation and Comparison of NLP Systems
%D 2021
%8 November
%I Association for Computational Linguistics
%C Punta Cana, Dominican Republic
%F treviso-etal-2021-ist
%X We present the joint contribution of Instituto Superior Técnico (IST) and Unbabel to the Explainable Quality Estimation (QE) shared task, where systems were submitted to two tracks: constrained (without word-level supervision) and unconstrained (with word-level supervision). For the constrained track, we experimented with several explainability methods to extract the relevance of input tokens from sentence-level QE models built on top of multilingual pre-trained transformers. Among the different tested methods, composing explanations in the form of attention weights scaled by the norm of value vectors yielded the best results. When word-level labels are used during training, our best results were obtained by using word-level predicted probabilities. We further improve the performance of our methods on the two tracks by ensembling explanation scores extracted from models trained with different pre-trained transformers, achieving strong results for in-domain and zero-shot language pairs.
%R 10.18653/v1/2021.eval4nlp-1.14
%U https://aclanthology.org/2021.eval4nlp-1.14
%U https://doi.org/10.18653/v1/2021.eval4nlp-1.14
%P 133-145
Markdown (Informal)
[IST-Unbabel 2021 Submission for the Explainable Quality Estimation Shared Task](https://aclanthology.org/2021.eval4nlp-1.14) (Treviso et al., Eval4NLP 2021)
ACL