Intend to analyze Social Media feeds to detect behavioral trends of individuals to proactively act against Social Threats

Just few years ago, many of us and rest of the world believed that social media was just fun, unproductive and pointless technical whim. But today, the use of social media has become a necessity and needless to mention, the crucial part of everyone's life, career & business. Today with the use of social media, people connect with people from whole world, express themselves, advertise their stuff & brands, and in many other forms. Although, Social Media is not just a business tool or a tool for self-promotion, it can surely contribute in a much larger spectrum and intensity to the society. Social media can help predict a Crime and its prevention too! To prevent events of crime by analyzing the behavioral changes and sentiment analysis of a person by predicting his/her thought process and line of thoughts through his/her social media is possible with the use of readily available and reliable techniques like Iterative Clustering. It can be used to provide immediate help and solution (with a collaborate team of Law Enforcement Agencies and Mental Health professionals/ Psychoanalysts) to a person who is going to risk his own life (suicide), or who is going to be a part of such criminal act. As a proactive measure, it can help provide immediate services/information to the victims and people in danger too. In this paper, an algorithm is proposed to find the person who are deviating from their normal behavior using unsupervised learning(clustering) and further classifying them in predefined categories. If the behavior change is within first threshold, it shows the change in behavior is normal and may not harm anyone. If the change is above first threshold, it shows noticeable change in behavior and need some attention by friends and family. But if the change is above second threshold value, that shows drastic change in behavior and may require immediate action. Profile of this person could also be verified by data provided with the police if any previous crime record is available with them.