Topic-segmentation corpora Arabic Corpora used in topic segmentation. Corpora composed of 3 datasets: 1- Multiparty conversation transcripts. 2- Written texts: Civilization story. 3- Synthetic dataset collected from OSAC Corpus.