Toward an automatic summarisation of Arabic text depending on rhetorical relations

Rhetorical relations between two text segments are crucial information and have been proven useful for many natural language processing applications. In this paper, we propose a supervised approach for automatic identifying of rhetorical relations in Arabic texts. Our model attempts to identify both implicit and explicit rhetorical relations between elementary discourse units which will be exploited in automatic summarisation of Arabic texts. To carry out this research, we developed a discourse annotated corpus following the rhetorical structure theory framework with high reliability. Relations annotation was done using a set of 23 fine-grained relations enriched with nuclearity annotation. To automatically learn these relations, we reuse some state of the arts features and contribute new lexical and semantics' features. The experimental results on fine-grained and coarse-grained relations show that our model achieved best performance relative to all baselines.