The Content Based Misinformation Detection for Gujarati Language

Misinformation detection (MID) has recently gained attention as a research question. We discover that many efforts have been made in response to the innovative research issues and MID research approaches as an empowering and developing rapidly research field. Misinformation typically spreads faster, profound, and wider in social networks. It is crucial to identify disinformation on social media since people are less able to distinguish between true information and false information due to an abundance of information and limited attention. Exploration of misinformation detection received little attention from the computational NLP research community. In this paper, we propose a novel architecture for detecting misinformation in Gujarati text. In the architecture, the significance of domain experts, crowd intelligence, and fact-checking website interaction has been depicted. Additionally, we define certain special Gujarati language traits for the early detection of counterfeit news. We have built a classifier incorporating domain experts to identify fake news and present outcomes for the automatic fake news detection and its class. We showed the findings of the experiments performed on the news article corpus that we built by fetching the articles from newspaper websites and applications. We found that stochastic Gradient Descent and NuSVC achieved the accuracy 88% using TFIDF and CV respectively, which is the highest among all models we experimented.