Hate Speech Classification Implementing NLP and CNN with Machine Learning Algorithm Through Interpretable Explainable AI

In recent years, data innovation has progressed quickly, and online media has gone through a problematic transformation. Destinations like Facebook, Twitter, and Instagram where clients can communicate their thoughts through messages, photographs, and recordings have become famous. Sadly, be that as it may, it has turned into a favorable place for disdain discourse, affronts, cyberbullying, and mysterious dangers. There are many examinations around here, however, none have given an adequate degree of exactness. This article proposes a Convolutional Neural Network (CNN) and Natural Language Processing (NLP) amalgamation strategy that characterizes malicious and non-malicious remarks at a beginning phase and groups them into six classifications utilizing Wikipedia's talk page edits gathered by Kaggle. The proposed engineering changes over words into words by utilizing NLP strategies like tokenization and stemming, and word embedding as well as data preprocessing methods that transform words into vectors. In the subsequent advance, the handled informational index gathered in the initial step is applied through 5 distinct classifiers (XGBoost, Random Forest, Decision Tree, AdaBoost, and Gradient Boosting classifier) and an explainable artificial intelligence (XAI) strategy (LIME). In the last stage, all classifiers were joined to accomplish a malicious comment classification accuracy of 99.75, which is higher than the past work.