The rapid dissemination of current events worldwide is enabled by social media. The simplicity of the social media platform makes it effective for the rapid dissemination of news and stories among its users. The majority of people consume and spread information without first giving serious consideration to whether it is accurate or not. The most effective way to alert users about falsehoods or polarization is to tag news with credibility indicators or send a user warning message. For generating credibility indicators (CI) to send warning messages and tagging, a Topic Model is proposed that extracts topical terms related to a named entity based on proximity, coherence, and popularity during a specific period. Topic coherence and quality metrics are used to assess the proposed Topic Model to Extract CI (TM-CI). The research paper describes a Fake News Detection model that utilizes the proposed Topic Model (TM-CI) in classifying the news content and generates warning messages to be sent to the user. The proposed Model based on TM-CI for Message (FND-CIM) uses a four-step model based on selecting a machine learning model based on its accuracy for a large-scale text corpus. The novel approach takes into account specific features like named entities, likes, comments, and shares on a user’s social media account. The performance evaluation parameters used to verify the findings were false positives, True negatives, recall, precision, accuracy, and F1-score.
Avasthi et al. (Sun,) studied this question.