<![CDATA[This study aims to develop an emotional Arabic corpus and to examine affection classification based on balāghah structures as an analytical framework. The primary objective is to provide a computational model capable of detecting and classifying emotions in Arabic texts while considering aspects of classical stylistics. The methodology involves the collection of Arabic textual data from both classical and contemporary literary sources, affect annotation conducted by linguists and experts in balāghah, and the application of machine learning algorithms for emotion classification. The analysis focuses on the relationship between stylistic devices—such as majāz, tashbīh, isti‘ārah, and other rhetorical patterns—and the emotional intensity represented in the corpus. The results indicate that balāghah structures contribute significantly to classification accuracy, particularly in detecting subtle and context-dependent emotional nuances. The developed classification model demonstrates superior performance compared to purely lexicon-based approaches, achieving an improvement in accuracy of approximately 12–15% in complex emotion categories. These findings highlight the relevance of traditional linguistic approaches in modern affective analysis and open opportunities for the development of Arabic natural language processing (NLP) applications that are sensitive to emotional and rhetorical contexts. In conclusion, the integration of balāghah structures into an emotional corpus can enhance computational understanding of the affective dimensions of the Arabic language while also providing a methodological contribution to the fields of linguistics and artificial intelligence.]]>
Copyrights © 2026
