Hate Speech Detection using CSO based Polynomial Network using Twitter

Abstract

The power of social media as a catalyst for societal transformation is now unrivalled. What happens in one part of the world has repercussions in other parts of the world. This is because the vast quantities of data produced by these platforms may be instantly disseminated to any part of the globe. To make cyber space as welcoming and productive as feasible for all users, developers of these platforms must overcome several obstacles. However, provocative speech and hate speech have emerged as major problems in recent years. The scale of this issue is so large that it cannot be solved by coordinated teamwork alone, no matter how hard people try. Actually, there is a need for the development of an automated approach that can identify and eliminate nasty and insulting remarks before they can do any damage. This paper offers a novel Deep Learning-based Hate Speech Detection Scheme (DL-HSDS) to identify hate speech in Twitter data. Even though there are a lot of HSDS methods available, many of them suffer from insufficient feature learning and poor dataset management, both of which negatively impact attack detection precision. Therefore, to improve detection accuracy, the suggested module integrates the Cuckoo Search Optimization algorithm (CSO) with the (SDPN); CSO picks the optimum features in the datasets, and SDPN categorises the data as hate or normal. The suggested model, which employs the tweet text with CSO to imprisonment the tweets' outperforms the previous models.