Overview of the proposed semi-supervised learning method for emotion detection in tweets: i) the dataset is divided into unannotated tweets and manually annotated tweets;...
Abstract:
This study proposes a novel semi-supervised multi-label emotion classification approach for French tweets based on pseudo-labeling. Human subjectivity in emotional expres...Show MoreMetadata
Abstract:
This study proposes a novel semi-supervised multi-label emotion classification approach for French tweets based on pseudo-labeling. Human subjectivity in emotional expression makes it difficult for a machine to learn. Therefore, it necessitates training supervised learning models on large datasets annotated by multiple annotators. However, creating such datasets can be costly and time-consuming. Moreover, aggregating annotations from multiple annotators to capture their collective emotional state adds complexity to the task. Semi-supervised learning techniques have shown effectiveness with limited datasets. Furthermore, Large language Models (LLMs), particularly Chat-GPT, have demonstrated superior annotation accuracy compared to annotations obtained from crowdsourcing platforms, when both are evaluated against expert-annotated data. This work introduces a novel approach for multi-label emotion classification of French tweets by leveraging pseudo-labels generated through Chat-GPT, a robust large language model. Using zero-shot, one-shot, and few-shot learning techniques, Chat-GPT annotates the unlabelled part of our dataset. These Chat-GPT-annotated pseudo-labels are then merged with manual annotations, facilitating the training of a multi-label emotion classification model via semi-supervised learning. Furthermore, within the context of our research, we present a new French tweet dataset, containing testimonials from people affected by an urban industrial incident. This dataset features 2,350 tweets, each manually annotated by three human annotators based on 8 pre-identified emotions. Benchmark results are presented for multi-label emotion classification models employing both fully supervised and semi-supervised approaches with pseudo-labeling. Our findings demonstrate the superiority of our approach for multi-label emotion classification over standard pseudo-labeling and an established baseline.
Overview of the proposed semi-supervised learning method for emotion detection in tweets: i) the dataset is divided into unannotated tweets and manually annotated tweets;...
Published in: IEEE Access ( Volume: 12)
Funding Agency:

Université de Rouen Normandie, LITIS UR 4108, Rouen, France
Usman Malik received the Ph.D. degree from Normandie Université, France, where his research focused on the application of machine learning and deep learning techniques to enhance multi-model human-agent interaction. Currently, he is actively engaged in the CATCH project, where his focus lies in the development of multi-label emotion classification models tailored for French-language tweets.
Usman Malik received the Ph.D. degree from Normandie Université, France, where his research focused on the application of machine learning and deep learning techniques to enhance multi-model human-agent interaction. Currently, he is actively engaged in the CATCH project, where his focus lies in the development of multi-label emotion classification models tailored for French-language tweets.View more

Université de Rouen Normandie, LITIS UR 4108, Rouen, France
Simon Bernard received the Ph.D. degree in computer science from the University of Rouen Normandy, France, in 2009. He has been an Associate Professor (Maitre de Conféences) with the University of Rouen, Normandy, since 2013. He is also a member of the LITIS Laboratory and Normastic CNRS Research Federation. His research interests include machine learning, ensemble learning, and deep learning, in various learning contexts...Show More
Simon Bernard received the Ph.D. degree in computer science from the University of Rouen Normandy, France, in 2009. He has been an Associate Professor (Maitre de Conféences) with the University of Rouen, Normandy, since 2013. He is also a member of the LITIS Laboratory and Normastic CNRS Research Federation. His research interests include machine learning, ensemble learning, and deep learning, in various learning contexts...View more

INSA Rouen Normandie, LITIS UR 4108, Rouen, France
Alexandre Pauchet is currently an Associate Professor with INSA Rouen Normandy, France, and the LITIS Laboratory. He is also in charge of the MIND Team and a Co-Animator of the Data, Learning and Knowledge Axis of the Normastic Federation. He has defended the French ability to supervise researches (HDR) with the University of Rouen, in 2015. His research interests include human-agent interaction, affective computing, and ...Show More
Alexandre Pauchet is currently an Associate Professor with INSA Rouen Normandy, France, and the LITIS Laboratory. He is also in charge of the MIND Team and a Co-Animator of the Data, Learning and Knowledge Axis of the Normastic Federation. He has defended the French ability to supervise researches (HDR) with the University of Rouen, in 2015. His research interests include human-agent interaction, affective computing, and ...View more

INSA Rouen Normandie, LITIS UR 4108, Rouen, France
Clément Chatelain is currently an Associate Professor with the Department of Information Systems Engineering, INSA Rouen Normandy, France. His research interests include machine learning applied to handwriting recognition, document image analysis, and medical image analysis. His teaching and research interests include signal processing, deep learning, and pattern recognition. In 2019, he received the French ability to sup...Show More
Clément Chatelain is currently an Associate Professor with the Department of Information Systems Engineering, INSA Rouen Normandy, France. His research interests include machine learning applied to handwriting recognition, document image analysis, and medical image analysis. His teaching and research interests include signal processing, deep learning, and pattern recognition. In 2019, he received the French ability to sup...View more

Saagie, Rouen, France
Romain Picot-Clémente received the Ph.D. degree in computer science from the University of Burgundy, France, in 2011. He has been the Head of Artificial Intelligence with Saagie, since 2018. He is currently the Co-Director of the Joint Laboratory L-LiSa between LITIS and Saagie. His research interests include deep learning, weakly supervised learning applied to text, and time series data.
Romain Picot-Clémente received the Ph.D. degree in computer science from the University of Burgundy, France, in 2011. He has been the Head of Artificial Intelligence with Saagie, since 2018. He is currently the Co-Director of the Joint Laboratory L-LiSa between LITIS and Saagie. His research interests include deep learning, weakly supervised learning applied to text, and time series data.View more

Atmo Normandie, Rouen, France
Jérôme Cortinovis received the Ph.D. degree in physico-chemical modeling of the atmosphere from CNRS-Laboratoire d’Aérologie, Toulouse, in 2004. He has been an Innovation and Partnership Engineer with Atmo Normandie, since 2004. He has worked in particular on the development of numerical modelling of air quality and on the implementation of open data. He is currently a Coordinator of Incub’air, Atmo Normandie’s Innovation...Show More
Jérôme Cortinovis received the Ph.D. degree in physico-chemical modeling of the atmosphere from CNRS-Laboratoire d’Aérologie, Toulouse, in 2004. He has been an Innovation and Partnership Engineer with Atmo Normandie, since 2004. He has worked in particular on the development of numerical modelling of air quality and on the implementation of open data. He is currently a Coordinator of Incub’air, Atmo Normandie’s Innovation...View more

Université de Rouen Normandie, LITIS UR 4108, Rouen, France
Usman Malik received the Ph.D. degree from Normandie Université, France, where his research focused on the application of machine learning and deep learning techniques to enhance multi-model human-agent interaction. Currently, he is actively engaged in the CATCH project, where his focus lies in the development of multi-label emotion classification models tailored for French-language tweets.
Usman Malik received the Ph.D. degree from Normandie Université, France, where his research focused on the application of machine learning and deep learning techniques to enhance multi-model human-agent interaction. Currently, he is actively engaged in the CATCH project, where his focus lies in the development of multi-label emotion classification models tailored for French-language tweets.View more

Université de Rouen Normandie, LITIS UR 4108, Rouen, France
Simon Bernard received the Ph.D. degree in computer science from the University of Rouen Normandy, France, in 2009. He has been an Associate Professor (Maitre de Conféences) with the University of Rouen, Normandy, since 2013. He is also a member of the LITIS Laboratory and Normastic CNRS Research Federation. His research interests include machine learning, ensemble learning, and deep learning, in various learning contexts, such as weakly supervised learning, one-class classification and/or high-dimensional, and low sample size (HDLSS) classification.
Simon Bernard received the Ph.D. degree in computer science from the University of Rouen Normandy, France, in 2009. He has been an Associate Professor (Maitre de Conféences) with the University of Rouen, Normandy, since 2013. He is also a member of the LITIS Laboratory and Normastic CNRS Research Federation. His research interests include machine learning, ensemble learning, and deep learning, in various learning contexts, such as weakly supervised learning, one-class classification and/or high-dimensional, and low sample size (HDLSS) classification.View more

INSA Rouen Normandie, LITIS UR 4108, Rouen, France
Alexandre Pauchet is currently an Associate Professor with INSA Rouen Normandy, France, and the LITIS Laboratory. He is also in charge of the MIND Team and a Co-Animator of the Data, Learning and Knowledge Axis of the Normastic Federation. He has defended the French ability to supervise researches (HDR) with the University of Rouen, in 2015. His research interests include human-agent interaction, affective computing, and dialogue.
Alexandre Pauchet is currently an Associate Professor with INSA Rouen Normandy, France, and the LITIS Laboratory. He is also in charge of the MIND Team and a Co-Animator of the Data, Learning and Knowledge Axis of the Normastic Federation. He has defended the French ability to supervise researches (HDR) with the University of Rouen, in 2015. His research interests include human-agent interaction, affective computing, and dialogue.View more

INSA Rouen Normandie, LITIS UR 4108, Rouen, France
Clément Chatelain is currently an Associate Professor with the Department of Information Systems Engineering, INSA Rouen Normandy, France. His research interests include machine learning applied to handwriting recognition, document image analysis, and medical image analysis. His teaching and research interests include signal processing, deep learning, and pattern recognition. In 2019, he received the French ability to supervise researches from the University of Rouen.
Clément Chatelain is currently an Associate Professor with the Department of Information Systems Engineering, INSA Rouen Normandy, France. His research interests include machine learning applied to handwriting recognition, document image analysis, and medical image analysis. His teaching and research interests include signal processing, deep learning, and pattern recognition. In 2019, he received the French ability to supervise researches from the University of Rouen.View more

Saagie, Rouen, France
Romain Picot-Clémente received the Ph.D. degree in computer science from the University of Burgundy, France, in 2011. He has been the Head of Artificial Intelligence with Saagie, since 2018. He is currently the Co-Director of the Joint Laboratory L-LiSa between LITIS and Saagie. His research interests include deep learning, weakly supervised learning applied to text, and time series data.
Romain Picot-Clémente received the Ph.D. degree in computer science from the University of Burgundy, France, in 2011. He has been the Head of Artificial Intelligence with Saagie, since 2018. He is currently the Co-Director of the Joint Laboratory L-LiSa between LITIS and Saagie. His research interests include deep learning, weakly supervised learning applied to text, and time series data.View more

Atmo Normandie, Rouen, France
Jérôme Cortinovis received the Ph.D. degree in physico-chemical modeling of the atmosphere from CNRS-Laboratoire d’Aérologie, Toulouse, in 2004. He has been an Innovation and Partnership Engineer with Atmo Normandie, since 2004. He has worked in particular on the development of numerical modelling of air quality and on the implementation of open data. He is currently a Coordinator of Incub’air, Atmo Normandie’s Innovation Laboratory, which aims to test and disseminate innovative solutions for air quality (including odor issues).
Jérôme Cortinovis received the Ph.D. degree in physico-chemical modeling of the atmosphere from CNRS-Laboratoire d’Aérologie, Toulouse, in 2004. He has been an Innovation and Partnership Engineer with Atmo Normandie, since 2004. He has worked in particular on the development of numerical modelling of air quality and on the implementation of open data. He is currently a Coordinator of Incub’air, Atmo Normandie’s Innovation Laboratory, which aims to test and disseminate innovative solutions for air quality (including odor issues).View more