Title Of Paper:
A New Method for Text Segmentation in Persian Based on Lexical Cohesion
Author's Name :  SelmaMokhtar Zadehshahraki, MashallahAbbasiDezfouli
KeyWords:   Latent Semantic Analysis, Text Segmentation in Persian, Unsupervised Algorithm, Untrained Segmentation, Persian tiling Algorithm, Evaluation Criteria
Pages:  12 -17
Volume: 3
Issue: 10
Year: 2015




In this paper, we present a new segmentation algorithm based on Latent Semantic Analysis for segmentation of Persian texts. The presented algorithm is fully automatic, without training and based on lexical cohesion and it performs segmentation using semantic relationship between the blocks. Evaluation of results shows that our algorithm acts better than unsupervised Persian tiling segmentation method and F_measure with70.97% had a significant improvement.

Full Text: