We present an overview of the AraGenEval shared task, organized as part of the ArabicNLP 2025 conference. This task introduced the first benchmark suite for Arabic authorship analysis, featuring three subtasks: Authorship Style Transfer, Authorship Identification, and AI-Generated Text Detection. We curated high-quality datasets, including over 47,000 paragraphs from 21 authors and a balanced corpus of human- and AI-generated texts. The task attracted significant global participation, with 72 registered teams from 16 countries. The results highlight the effectiveness of transformer-based models, with top systems leveraging prompt engineering for style transfer, model ensembling for authorship identification, and a mix of multilingual and Arabic-specific models for AI text detection. This paper details the task design, datasets, participant systems, and key findings, establishing a foundation for future research in Arabic stylistics and trustworthy NLP.

This paper presents our system and findings for SemEval 2024 Task 1 Track A Supervised Semantic Textual Relatedness. The main objective of this task was to detect the degree of semantic relatedness between pairs of sentences. Our submitted models (ranked 6/24 in Algerian Arabic, 7/25 in Spanish, 12/23 in Moroccan Arabic, and 13/36 in English) consist of various transformer-based models including MARBERT-V2, mDeBERTa-V3-Base, DarijaBERT, and DeBERTa-V3-Large, fine-tuned using different loss functions including Huber Loss, Mean Absolute Error, and Mean Squared Error.

In this paper, we introduce our participating system to the ArAIEval Shared Task, addressing both the detection of persuasion techniques and disinformation tasks. Our proposed system employs a pre-trained transformer-based language model for Arabic, alongside a classifier. We have assessed the performance of three Arabic Pre-trained Language Models (PLMs) for sentence encoding. Additionally, to enhance our model’s performance, we have explored various training objectives, including Cross-Entropy loss, regularized Mixup loss, asymmetric multi-label loss, and Focal Tversky loss. On the official test set, our system has achieved micro-F1 scores of 0.7515, 0.5666, 0.904, and 0.8333 for Sub-Task 1A, Sub-Task 1B, Sub-Task 2A, and Sub-Task 2B, respectively. Furthermore, our system has secured the 4th, 1st, 3rd, and 2nd positions, respectively, among all participating systems in sub-tasks 1A, 1B, 2A, and 2B of the ArAIEval shared task.

In this paper, we present our submitted system for the WojoodNER Shared Task, addressing both flat and nested Arabic Named Entity Recognition (NER). Our system is based on a BERT-based multi-task learning model that leverages the existing Arabic Pretrained Language Models (PLMs) to encode the input sentences. To enhance the performance of our model, we have employed a multi-task loss variance penalty and combined several training objectives, including the Cross-Entropy loss, the Dice loss, the Tversky loss, and the Focal loss. Besides, we have studied the performance of three existing Arabic PLMs for sentence encoding. On the official test set, our system has obtained a micro-F1 score of 0.9113 and 0.9303 for Flat (Sub-Task 1) and Nested (Sub-Task 2) NER, respectively. It has been ranked in the 6th and the 2nd positions among all participating systems in Sub-Task 1 and Sub-Task 2, respectively.

This paper presents our system and findings for SemEval 2023 Task 9 Tweet Intimacy Analysis. The main objective of this task was to predict the intimacy of tweets in 10 languages. Our submitted model (ranked 28/45) consists of a transformer-based approach with data augmentation via machine translation.

This paper introduces our participating system to the Explainable Detection of Online Sexism (EDOS) SemEval-2023 - Task 10: Explainable Detection of Online Sexism. The EDOS shared task covers three hierarchical sub-tasks for sexism detection, coarse-grained and fine-grained categorization. We have investigated both single-task and multi-task learning based on RoBERTa transformer-based language models. For improving the results, we have performed further pre-training of RoBERTa on the provided unlabeled data. Besides, we have employed a small sample of the unlabeled data for semi-supervised learning using the minimum class-confusion loss. Our system has achieved macro F1 scores of 82.25\%, 67.35\%, and 49.8\% on Tasks A, B, and C, respectively.

This paper presents our proposed method for english documents genre classification in the context of SemEval 2023 task 3, subtask 1. Our method use ensemble technique to combine four distinct models predictions: Longformer, RoBERTa, GCN, and a sentences number-based model. Each model is optimized on simple objectives and easy to grasp. We provide snippets of code that define each model to make the reading experience better. Our method ranked 12th in documents genre classification for english texts.

This paper presents our submitted system to AfriSenti SemEval-2023 Task 12: Sentiment Analysis for African Languages. The AfriSenti consists of three different tasks, covering monolingual, multilingual, and zero-shot sentiment analysis scenarios for African languages. To improve model generalization, we have explored the following steps: 1) further pre-training of the AfroXLM Pre-trained Language Model (PLM), 2) combining AfroXLM and MARBERT PLMs using a residual layer, and 3) studying the impact of metric learning and two out-of-distribution generalization training objectives. The overall evaluation results show that our system has achieved promising results on several sub-tasks of Task A. For Tasks B and C, our system is ranked among the top six participating systems.

This paper presents our proposed methods for the iSarcasmEval shared task. The shared task consists of three different subtasks. We participate in both subtask A and subtask C. The purpose of subtask A was to predict if a text is sarcastic while the aim of subtask C is to determine which text is sarcastic given a sarcastic text and its non-sarcastic rephrase. Both of the developed solutions used BERT pre-trained models. The proposed models are optimized on simple objectives and are easy to grasp. However, despite their simplicity, our methods ranked 4 and 2 in iSarcasmEval subtask A and subtask C for Arabic texts.

In this paper, we present our system and findings for SemEval-2022 Task 6 - iSarcasmEval: Intended Sarcasm Detection in English. The main objective of this task was to identify sarcastic tweets. This task was challenging mainly due to (1) the small training dataset that contains only 3468 tweets and (2) the imbalanced class distribution (25% sarcastic and 75% non-sarcastic). Our submitted model (ranked eighth on Sub-Task A and fifth on Sub-Task C) consists of a Transformer-based approach (BERTweet model).

Toxic spans detection is an emerging challenge that aims to find toxic spans within a toxic text. In this paper, we describe our solutions to tackle toxic spans detection. The first solution, which follows a supervised approach, is based on SpanBERT model. This latter is intended to better embed and predict spans of text. The second solution, which adopts an unsupervised approach, combines linear support vector machine with the Local Interpretable Model-Agnostic Explanations (LIME). This last is used to interpret predictions of learning-based models. Our supervised model outperformed the unsupervised model and achieved the f-score of 67,84% (ranked 22/85) in Task 5 at SemEval-2021: Toxic Spans Detection.

AraBERT is an Arabic version of the state-of-the-art Bidirectional Encoder Representations from Transformers (BERT) model. The latter has achieved good performance in a variety of Natural Language Processing (NLP) tasks. In this paper, we propose an effective AraBERT embeddings-based method for dealing with offensive Arabic language in Twitter. First, we pre-process tweets by handling emojis and including their Arabic meanings. To overcome the pretrain-finetune discrepancy, we substitute each detected emojis by the special token [MASK] into both fine tuning and inference phases. Then, we represent tweets tokens by applying AraBERT model. Finally, we feed the tweet representation into a sigmoid function to decide whether a tweet is offensive or not. The proposed method achieved the best results on OffensEval 2020: Arabic task and reached a macro F1 score equal to 90.17%.

Around the Arab world, different Arabic dialects are spoken by more than 300M persons, and are increasingly popular in social media texts. However, Arabic dialects are considered to be low-resource languages, limiting the development of machine-learning based systems for these dialects. In this paper, we investigate the Arabic dialect identification task, from two perspectives: country-level dialect identification from 21 Arab countries, and province-level dialect identification from 100 provinces. We introduce an unified pipeline of state-of-the-art models, that can handle the two subtasks. Our experimental studies applied to the NADI shared task, show promising results both at the country-level (F1-score of 25.99%) and the province-level (F1-score of 6.39%), and thus allow us to be ranked 2nd for the country-level subtask, and 1st in the province-level subtask.