Eliminating Toxicity in Text: An NLP Framework for Clean Content Extraction

This research study proposes a machine learning-based application called ToxiGuard, which is designed to ascertain the toxic language in text. The proposed technique employs the use of the Natural Language Processing (NLP) mechanism and a Naive Bayes based classifier to detect and categorize unhealthy or toxic comments. First the given text data is preprocessed by dividing or tokenizing the sentences into different words, once tokens are formed it will be passed as an argument into the proposed method which removes all the words which are unhealthy based on the trained model. Model takes the decision based on the patterns correlated with toxic or unhealthy language. The proposed mechanism is represented in the form of a graphical user interface (GUI) built with Tkinter. GUI being designed enables user to upload a text file which contains serious of text or sentences analysis is being done and visual representation of the outcome is plotted with help of a bar chart of the maximum frequency of each toxic words. In addition to analysis report outcome of the analysis can be exported for further review process. ToxiGuard focus to help or classify online content by providing effective and user-friendly application in-order to detect harmful or unhealthy language.