Language & Media
Morteza Dastlan; Mahshid Dehbozorgi; Behzad Moridi; Mohsen Emami
Abstract
The distribution of fake sport news is not based on the satisfaction of sport men, sport clubs and sport fans. Correspondingly, the identification of fake news is important and practical. This research has been done in the framework of computational linguistics. The linguistic data are based on a corpus ...
Read More
The distribution of fake sport news is not based on the satisfaction of sport men, sport clubs and sport fans. Correspondingly, the identification of fake news is important and practical. This research has been done in the framework of computational linguistics. The linguistic data are based on a corpus of sports news from ISNA website and Instagram program. In this way, sports news is downloaded from the ISNA website in a period of time, and then in a few pages of the Instagram program, sports news is downloaded and compared in terms of being fake or not. The N-gram method and long and short term memory (LSTM) method have been used to identify fake news from non-fake ones. The method proposed in this paper has been implemented on four valid and existing datasets and has been compared with the previous six methods. The accuracy of this method is acceptable compared to other methods, and the results obtained indicate that this method is suitable and accurate enough to identify fake news among the news published on Instagram.
