Political Security Threat Prediction Framework Using Hybrid Lexicon-Based Approach and Machine Learning Technique

The internet provides a strong medium for the expression of opinions, emotions and ideas, through online media backed by smartphone use and high internet penetration. The majority of internet postings are text-based and may encompass people's emotional sentiments for a specific moment or feeling. Online sentiments or opinions should be monitored to identify any violent emotions evoked by citizens that can cause harm. Riots and warfare, for example, need to be tackled because there is a threat of undermining social stability and political security, which are very essential components of nations security. Mining opinions for national security is a pertinent research issue that needs to be developed further. Mechanisms and methods capable of mining opinions in the area of political security need to be greatly improved to yield optimal results. Academics have seen that there is also an intimate connection between emotion, sentiment and political security breaks. This research posits a new theory framework for forecasting political security threats based on a hybrid method: the integration of lexicon-based methods and machine learning in cyberspace. In the suggested framework, Decision Tree, Naive Bayes, and Support Vector Machine have been used as threat classifiers. To confirm our suggested framework, an experimental analysis is achieved. The performance of every technique employed in the experiments is presented. In this research, our suggested framework indicates that the hybrid Lexicon-based method with the RandomForest classifier achieved the highest performance score for political security threat prediction. Results show current research on opinion mining in threat prediction based on the political security domain.