A Novel Stance Labeling Algorithm for Misinformation Detection

With the ease with which the data is accessible over the Internet, the spread of ubiquitous information is increasing rapidly, resulting in long-lasting effects on people, businesses, the economy, and so on. Detecting and mitigating the misinformation generated online is still challenging due to its rapid generation. Researchers have developed models that detect and classify misinformation by extracting numerous features from the textual contents and defining user-specific features. One of the important features is a stance, which articulates the viewpoint of the speaker and assists in determining the truthfulness of the articles. Recent literature does manual labeling of stance and true and false categorization of information, making it an uphill task. Therefore, in this research, authors have proposed a novel algorithm that can classify the stance automatically into two categories: support and denial. Further, these features are used to build models to classify news articles as true and false. With this approach, the model showed the highest accuracy of 85.26%.