CS-UM6P at SemEval-2022 Task 6: Transformer-based Models for Intended Sarcasm Detection in English and Arabic

Mahdaouy, Abdelkader El; Mekki, Abdellah El; Essefar, Kabil; Skiredj, Abderrahman; Berrada, Ismaïl

doi:10.18653/v1/2022.semeval-1.117

Cited by 5 publications

(6 citation statements)

References 12 publications

(23 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…For NestedNER, the ELYADATA team (Laouirine et al, 2023) ranks in the first position with an F 1 score of 93.73, followed by UM6P & UL team (El Mahdaouy et al, 2023) with a score of 93.09 and in third place AlexU-AIC with a score of 92.61. Notably, there are four teams that outperform baseline-I with F 1 score gap between the baseline and the best model of 2.05%.…”

Section: Resultsmentioning

confidence: 99%

“…The highest performance for the product is 66.67% which is obtained by ThinkNER team. For the quantity, the 63.16% F1-score is obtained by (El Mahdaouy et al, 2023). For website, the best performance is 69.26% F1-score.…”

Section: Resultsmentioning

confidence: 99%

“…For instance, Elyadata finetuned BioBERT (Lee et al, 2020), but the results were much worse than the baseline, which is expected since BioBERT is trained on English biomedical corpus. In a comparative study, the UM6P & UL (El Mahdaouy et al, 2023) team explored the capabilities of QARiB (Abdelali et al, 2021), a model pre-trained specifically on Arabic tweets, against ARBERT v2 (Abdul-Mageed et al, 2021), which is trained on an expansive and diverse Arabic datasets. Their finding shows ARBERT v2 's superiority over other models.…”

Section: General Description Of Submitted Systemsmentioning

confidence: 99%

See 2 more Smart Citations

WojoodNER 2023: The First Arabic Named Entity Recognition Shared Task

Jarrar,

Abdul-Mageed,

Khalilia

et al. 2023

Proceedings of ArabicNLP 2023

View full text Add to dashboard Cite

We present WojoodNER-2023, the first Arabic Named Entity Recognition (NER) Shared Task. The primary focus of WojoodNER 2023 is on Arabic NER, offering novel NER datasets (i.e., Wojood) and the definition of subtasks designed to facilitate meaningful comparisons between different NER approaches. WojoodNER-2023 encompassed two Subtasks: FlatNER and NestedNER. A total of 45 unique teams registered for this shared task, with 11 of them actively participating in the test phase. Specifically, 11 teams participated in FlatNER, while 8 teams tackled NestedNER. The winning teams achieved F 1 scores of 91.96 and 93.73 in Flat-NER and NestedNER, respectively.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: General Description Of Submitted Systemsmentioning

confidence: 99%

See 1 more Smart Citation

WojoodNER 2023: The First Arabic Named Entity Recognition Shared Task

Jarrar,

Abdul-Mageed,

Khalilia

et al. 2023

Proceedings of ArabicNLP 2023

View full text Add to dashboard Cite

show abstract

“…Several approaches are adopted to enhance the base performance of BERT for sentiment analysis and sarcasm detection. Mahdaouy et al (2021) added an attention layer on top of MARBERT and constructed the sentence embedding by concatenating the [CLS] embedding with the output of the attention layer. They also tackled multitask learning through the same architecture by adding an extra attention layer for the other task.…”

Section: Related Workmentioning

confidence: 99%

Rethinking of BERT Sentence Embedding for Text Classification

Galal,

Abdel-Gawad,

Farouk

2024

Preprint

View full text Add to dashboard Cite

Text classification is a fundamental task in NLP that is used in several real-life tasks and applications. Large pre-trained language models such as BERT achieve state-of-the-art performance in several NLP tasks including text classification tasks. Although BERT boosts text classification performance, the common way of using it for classification lacks many aspects of its advantages. This work rethinks the way of using BERT final layer and hidden layers embeddings by proposing different aggregation architectures for text classification tasks such as sentiment analysis and sarcasm detection. This research also proposes different approaches for using BERT as a feature extractor without fine-tuning whose performance surpasses its fine-tuning counterpart. It also proposes promising multi-task learning aggregation architectures to improve the performance of the related classification problems. The experiments of the different architectures show that freezing BERT can outperform fine-tuning it for sentiment analysis. The experiments also show that multi-task learning while freezing BERT boosts the performance of yet hard tasks such as sarcasm detection. The best-performing models achieved new state-of-the-art performance on the ArSarcasm-v2 dataset for Arabic sarcasm detection and sentiment analysis. For multi-task learning and freezing BERT, a new SOTA F1-score of 64.41 was achieved for the sarcasm detection with a 3.47% improvement and near SOTA FPN of 75.78 for the sentiment classification. For single-task learning, a new SOTA FPN of 75.26 was achieved for the sentiment with a 1.81% improvement.

show abstract

“…На сегодняшний день, наиболее высокое качество во многих задачах классификации текстов показывают нейросетевые модели, основанные на архитектуре Transformer [28] и, в частности, на использовании лингвистических моделей Bidirectional Encoder Representations from Transformers (BERT) [29], RoBERTa [30] и их модификаций. Так, в ряде соревнований по машинному обучению, связанных с тематикой классификации текстов социальных сетей и проводимых в рамках крупнейших конференций, лучшие результаты были получены с помощью BERT и ее вариаций (например, [31][32][33]).…”

Section: обзор смежных работunclassified

Detecting Mentions of Green Practices in Social Media Based on Text Classification

Glazkova

Zakharova

Zakharov

et al. 2022

Model. anal. inf. sist.

View full text Add to dashboard Cite

The paper is devoted to the task of searching for mentions of green practices in social media texts. The relevance of this task is dictated by the need to expand existing knowledge about the use of green practices in society and the spread of existing green practices. This paper uses a text corpus consisting of the texts published on the environmental communities of the VKontakte social network. The corpus is equipped with an expert markup of the mention of nine types of green practices. As part of this work, a semi-automatic approach is proposed to the collection of additional texts to reduce the class imbalance in the corpus. The approach includes the following steps: detecting the most frequent words for each practice type; automatic collecting texts in social media that contain the detected frequent words; expert verification and filtering of collected texts. The four machine learning models are compared to find the mentions of green practices on the two variants of the corpus: original and augmented using the proposed approach. Among the listed models, the highest averaged F1-score (81.32%) was achieved by Conversational RuBERT fine-tuned on the augmented corpus. Conversational RuBERT model was chosen for the implementation of the application prototype. The main function of the prototype is to detect the presence of the mention of nine types of green practices in the text. The prototype is implemented in the form of the Telegram chatbot.

show abstract

CS-UM6P at SemEval-2022 Task 6: Transformer-based Models for Intended Sarcasm Detection in English and Arabic

Cited by 5 publications

References 12 publications

WojoodNER 2023: The First Arabic Named Entity Recognition Shared Task

WojoodNER 2023: The First Arabic Named Entity Recognition Shared Task

Rethinking of BERT Sentence Embedding for Text Classification

Detecting Mentions of Green Practices in Social Media Based on Text Classification

Contact Info

Product

Resources

About