Abstract:

The emergence of the World Wide Web and the swift embrace of social media platforms, such as Facebook and Twitter, facilitated an unprecedented level of information distribution in human history. In addition to many applications, news organizations have gained from the extensive utilization of social media platforms by delivering timely news updates to its customers. The news media transitioned from traditional formats like newspapers, tabloids, and magazines to digital ones, including online news platforms, blogs, social media feeds, and various other digital media formats. Consumers found it increasingly convenient to access the newest news at their fingertips. Seventy percent of traffic to news websites originates from Facebook recommendations. The present iteration of these social media platforms is highly influential and beneficial, facilitating user discussions, idea sharing, and debates on topics such as democracy, education, and health. Nonetheless, these channels are also exploited by certain groups for financial profit, as well as for fostering prejudiced viewpoints, influencing perceptions, and disseminating satire or absurdity. This tendency is widely referred to as misinformation. The proliferation of misinformation has escalated significantly over the past decade, most evident during the 2016 US elections. The widespread dissemination of factually inaccurate material online has resulted in several issues, extending beyond politics to encompass different fields such as sports, health, and science. The financial markets are one sector impacted by misinformation, where a single rumor can lead to catastrophic outcomes and can disrupt market operations. Consequently, an automated method for the precise classification of authentic and fraudulent news is essential. Although some research have been completed, greater investigation and focus are necessary. The suggested initiative seeks to eradicate the dissemination of rumors and misinformation by facilitating the automated classification of news sources as credible or not. Initially, N-gram Feature Selection is employed to select the most pertinent features from the dataset. Subsequently, long short-term memory (LSTM) is employed to execute the categorization task.

Description:

.