Enhancing Arabic Poetic Structure Analysis through Visualization

Abstract

This study explores the potential of visual representation in understanding the structural elements of Arabic poetry, a subject of significant educational and research interest. Our objective is to make Arabic poetic works more accessible to readers of both Arabic and non-Arabic linguistic backgrounds by employing visualization, exploration, and analytical techniques. We transformed poetry texts into syllables, identified their metrical structures, segmented verses into patterns, and then converted these patterns into visual representations. Following this, we computed and visualized the dissimilarities between these images, and overlaid their differences. Our findings suggest that the positional patterns across a poem play a pivotal role in effective poetry clustering, as demonstrated by our newly computed metrics. The results of our clustering experiments showed a marked improvement over previous attempts, thereby providing new insights into the composition and structure of Arabic poetry. This study underscored the value of visual representation in enhancing our understanding of Arabic poetry.