@inproceedings{khered-etal-2022-building,
title = "Building an Ensemble of Transformer Models for {A}rabic Dialect Classification and Sentiment Analysis",
author = "Khered, Abdullah Salem and
Abdelhalim, Ingy Yasser Hassan Abdou and
Batista-Navarro, Riza",
editor = "Bouamor, Houda and
Al-Khalifa, Hend and
Darwish, Kareem and
Rambow, Owen and
Bougares, Fethi and
Abdelali, Ahmed and
Tomeh, Nadi and
Khalifa, Salam and
Zaghouani, Wajdi",
booktitle = "Proceedings of the Seventh Arabic Natural Language Processing Workshop (WANLP)",
month = dec,
year = "2022",
address = "Abu Dhabi, United Arab Emirates (Hybrid)",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.wanlp-1.53",
doi = "10.18653/v1/2022.wanlp-1.53",
pages = "479--484",
abstract = "In this paper, we describe the approaches we developed for the Nuanced Arabic Dialect Identification (NADI) 2022 shared task, which consists of two subtasks: the identification of country-level Arabic dialects and sentiment analysis. Our team, UniManc, developed approaches to the two subtasks which are underpinned by the same model: a pre-trained MARBERT language model. For Subtask 1, we applied undersampling to create versions of the training data with a balanced distribution across classes. For Subtask 2, we further trained the original MARBERT model for the masked language modelling objective using a NADI-provided dataset of unlabelled Arabic tweets. For each of the subtasks, a MARBERT model was fine-tuned for sequence classification, using different values for hyperparameters such as seed and learning rate. This resulted in multiple model variants, which formed the basis of an ensemble model for each subtask. Based on the official NADI evaluation, our ensemble model obtained a macro-F1-score of 26.863, ranking second overall in the first subtask. In the second subtask, our ensemble model also ranked second, obtaining a macro-F1-PN score (macro-averaged F1-score over the Positive and Negative classes) of 73.544.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="khered-etal-2022-building">
<titleInfo>
<title>Building an Ensemble of Transformer Models for Arabic Dialect Classification and Sentiment Analysis</title>
</titleInfo>
<name type="personal">
<namePart type="given">Abdullah</namePart>
<namePart type="given">Salem</namePart>
<namePart type="family">Khered</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ingy</namePart>
<namePart type="given">Yasser</namePart>
<namePart type="given">Hassan</namePart>
<namePart type="given">Abdou</namePart>
<namePart type="family">Abdelhalim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Riza</namePart>
<namePart type="family">Batista-Navarro</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Seventh Arabic Natural Language Processing Workshop (WANLP)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Houda</namePart>
<namePart type="family">Bouamor</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hend</namePart>
<namePart type="family">Al-Khalifa</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kareem</namePart>
<namePart type="family">Darwish</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Owen</namePart>
<namePart type="family">Rambow</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fethi</namePart>
<namePart type="family">Bougares</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ahmed</namePart>
<namePart type="family">Abdelali</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nadi</namePart>
<namePart type="family">Tomeh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Salam</namePart>
<namePart type="family">Khalifa</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wajdi</namePart>
<namePart type="family">Zaghouani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Abu Dhabi, United Arab Emirates (Hybrid)</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>In this paper, we describe the approaches we developed for the Nuanced Arabic Dialect Identification (NADI) 2022 shared task, which consists of two subtasks: the identification of country-level Arabic dialects and sentiment analysis. Our team, UniManc, developed approaches to the two subtasks which are underpinned by the same model: a pre-trained MARBERT language model. For Subtask 1, we applied undersampling to create versions of the training data with a balanced distribution across classes. For Subtask 2, we further trained the original MARBERT model for the masked language modelling objective using a NADI-provided dataset of unlabelled Arabic tweets. For each of the subtasks, a MARBERT model was fine-tuned for sequence classification, using different values for hyperparameters such as seed and learning rate. This resulted in multiple model variants, which formed the basis of an ensemble model for each subtask. Based on the official NADI evaluation, our ensemble model obtained a macro-F1-score of 26.863, ranking second overall in the first subtask. In the second subtask, our ensemble model also ranked second, obtaining a macro-F1-PN score (macro-averaged F1-score over the Positive and Negative classes) of 73.544.</abstract>
<identifier type="citekey">khered-etal-2022-building</identifier>
<identifier type="doi">10.18653/v1/2022.wanlp-1.53</identifier>
<location>
<url>https://aclanthology.org/2022.wanlp-1.53</url>
</location>
<part>
<date>2022-12</date>
<extent unit="page">
<start>479</start>
<end>484</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Building an Ensemble of Transformer Models for Arabic Dialect Classification and Sentiment Analysis
%A Khered, Abdullah Salem
%A Abdelhalim, Ingy Yasser Hassan Abdou
%A Batista-Navarro, Riza
%Y Bouamor, Houda
%Y Al-Khalifa, Hend
%Y Darwish, Kareem
%Y Rambow, Owen
%Y Bougares, Fethi
%Y Abdelali, Ahmed
%Y Tomeh, Nadi
%Y Khalifa, Salam
%Y Zaghouani, Wajdi
%S Proceedings of the Seventh Arabic Natural Language Processing Workshop (WANLP)
%D 2022
%8 December
%I Association for Computational Linguistics
%C Abu Dhabi, United Arab Emirates (Hybrid)
%F khered-etal-2022-building
%X In this paper, we describe the approaches we developed for the Nuanced Arabic Dialect Identification (NADI) 2022 shared task, which consists of two subtasks: the identification of country-level Arabic dialects and sentiment analysis. Our team, UniManc, developed approaches to the two subtasks which are underpinned by the same model: a pre-trained MARBERT language model. For Subtask 1, we applied undersampling to create versions of the training data with a balanced distribution across classes. For Subtask 2, we further trained the original MARBERT model for the masked language modelling objective using a NADI-provided dataset of unlabelled Arabic tweets. For each of the subtasks, a MARBERT model was fine-tuned for sequence classification, using different values for hyperparameters such as seed and learning rate. This resulted in multiple model variants, which formed the basis of an ensemble model for each subtask. Based on the official NADI evaluation, our ensemble model obtained a macro-F1-score of 26.863, ranking second overall in the first subtask. In the second subtask, our ensemble model also ranked second, obtaining a macro-F1-PN score (macro-averaged F1-score over the Positive and Negative classes) of 73.544.
%R 10.18653/v1/2022.wanlp-1.53
%U https://aclanthology.org/2022.wanlp-1.53
%U https://doi.org/10.18653/v1/2022.wanlp-1.53
%P 479-484
Markdown (Informal)
[Building an Ensemble of Transformer Models for Arabic Dialect Classification and Sentiment Analysis](https://aclanthology.org/2022.wanlp-1.53) (Khered et al., WANLP 2022)
ACL