Opinion mining hybrid technique to classify people's emotions in text using Kansei and lexicon-based approach for national security domain

Abstract

Current information sharing networks today offer a valuable medium to express emotions, feelings, ideas and opinions, as well as interactions between individuals from different walks of life. Nowadays, there are currently 2.62 billion social networks for active users. Social networks are used to exchange ideas and knowledge allowing individuals, groups and organisations to connect. This provides a large and rich pool of knowledge that can play a critical role for corporations, governments, political campaigns as well as administrative management and welfare in decision making. Most posts on the Internet are in the form of text, which can be processed to mine people's sentiments and emotional feelings. People do not realise that text on digital media platforms can be a threat to society's peace when judged with excessive emotions. Excessive emotions from citizens can bring unwanted risks including riots and civil war. Any threat to society's peace needs to be addressed as it can weaken national security. For this purpose, this paper explores a hybrid technique to classify people's emotions in text from online news that can affect national security using the lexicon-based approach and Kansei to determine sentiment polarity and emotions in text. The findings from this study can be utilised as a foundation for opinion mining in the national security domain.