Multimodal News Authentication As a Service

The «True News» Extension

Abstract

The current work focuses on the problem of misinformation. Filtering and blocking every unreliable source is impossible even to consider. Thus, discovering the dedicated steps to indicate fake content, according to the fact-checking procedures, and utilize them in automated/semi-automated mechanisms, is the key to defend the truth. Despite the availability of various authentication applications/services, there is a lack of integrated systems supporting media veracity in real-world scenarios. There are useful tools and practices for detecting processed/altered content, usually investigating a single manipulation or relying on the credibility of one source, which turns to be inadequate. Based on the above, a browser extension is presented, aiming at evaluating news authenticity in a multimodal, integrating and collaborative way. Operating unobtrusively in the background until needed, the extension is a solution transparent to the user. Without having to open a new browsing tab or to switch navigation /media environment, the user can identify relevant information regarding the five (5) clues, which frame a news story (Title, Date, Creator, Source, Containing Images). In essence, with the aid of the True News Plugin, a user reveals information from trusted sources, classifying them by the accuracy of their domain names, while also spotting possible misinformation through doctored images. The main novelty of the proposed module is that it will help users determine whether they can trust an article or not. By going through five (5) essential steps and by answering ten (10) questions, the proposed methodology attempts to introduce a valuable free tool in the field of Digital Forensics.