Scholarlink

This research addresses the challenge of classifying Arabic dialects into five main categories: GULF (Gulf Arabic), LEV (Levantine Arabic), NOAFTD and NOAF-MD (North African Arabic – Tunisian and Moroccan dialects), and EGY (Egyptian Arabic). We employ advanced techniques to extract meaningful linguistic features, such as TF-IDF, n-grams, CBOW, and Word2Vec, to capture the unique characteristics of each dialect. To handle the complexity of high-dimensional data, we use dimensionality reduction methods like Principal Component Analysis (PCA) and Singular Value Decomposition (SVD). The results highlight XGBoost as the top-performing model, achieving up to 92% accuracy, recall, and F1-score with TF-IDF and Unigram representations. These findings demonstrate the effectiveness of combining feature extraction techniques with advanced machine learning models for highly accurate Arabic dialect classification, setting a strong foundation for future NLP applications in multilingual settings.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Detecting Dialectal Variations in Arabic: A Comparative Study of North African, Levantine, and Gulf Arabic Using Extraction Feature Selection Techniques

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Abdellah Ait Elouli,
  • Hassan Ouahi,
  • El Mehdi Cherrat,
  • Abdellatif Bekkar

摘要

This research addresses the challenge of classifying Arabic dialects into five main categories: GULF (Gulf Arabic), LEV (Levantine Arabic), NOAFTD and NOAF-MD (North African Arabic – Tunisian and Moroccan dialects), and EGY (Egyptian Arabic). We employ advanced techniques to extract meaningful linguistic features, such as TF-IDF, n-grams, CBOW, and Word2Vec, to capture the unique characteristics of each dialect. To handle the complexity of high-dimensional data, we use dimensionality reduction methods like Principal Component Analysis (PCA) and Singular Value Decomposition (SVD). The results highlight XGBoost as the top-performing model, achieving up to 92% accuracy, recall, and F1-score with TF-IDF and Unigram representations. These findings demonstrate the effectiveness of combining feature extraction techniques with advanced machine learning models for highly accurate Arabic dialect classification, setting a strong foundation for future NLP applications in multilingual settings.