Machine learning approach for identification of threat content in audio messages shared on social media

Abstract

Now-a-days social media is on boom and it becomes a part of individuals’ daily life. All most all the people communicate with their friends, family members, peers etc using any one of the social media tools (e.g. Twitter, Snapchat, Facebook, and Instagram). People not only share their daily routine with others but sometimes share confidential information with each other in the forms of texts, images and audio messages. Depending on the existing political and economic instability in many countries, there is a need to monitor the content shared among users on social media. A lot of research has undergone in the past on how to monitor the shared content on social media, but they only address the text and images to some extent. According to researcher’s knowledge there exists no research which address to disclose the content in audio messages which are shared on social media. In this paper the researcher’s proposed a methodology which helps in detecting the unusual content shared among users on social media like (hazard, threat etc words) in the form of audio messages using Deep Learning. The results of this research are helpful in monitoring the social media content specially the audio messages by identifying hazards or threats planned before execution and in turn saving the society and mankind from destruction.

Subject Classification:

• 68T99

Keywords:

• Social Media
• Hazard
• Deep Learning
• TensorFlow
• Threat