Text Analysis Using Morphological Operations on a Neutrosophic Text Hypergraph

Abstract

Due to the rise in the operation of platforms on social media, there is more opportunity for users to post content online, out of which some tend to be hate speech. Hate speech is found in almost all domains like sports, politics, religion, government affairs, and personal matters. Its detection and removal from platforms like Twitter, Facebook, etc. are tedious. Over the years, a lot of methods have evolved in this area most of which are more time-consuming machine learning methods. Our objective is to find a better method that considers indeterminacy at the word level and sentence level for the detection and removal of hate speech using fuzzy logic applied to Neutrosophic hypergraphs. A neutrosophic hypergraph is a kind of hypergraph where each node and hyperedge has three associated membership functions namely Indeterminacy, Truth and Falsity. Our work has successfully modeled Text documents into neutrosophic hypergraphs and morphological operators like dilation, erosion etc. are applied to it. Using these operations further operators like thinning, thickening, hit-or-miss, and skeletoning are applied. Finally hate speech is identified and removed. This a novel method in this area. The system is tested with Twitter tweets and the results are promising with an accuracy of 88%.