NLP based Analysis and Detection of Unethical Text
Apurva A. Khandekar, Chekuri Devi Hema, Alle Meghana, Akuraju Mounika, V. S. Vaishnavi · 2023
In the present scenario, the internet has enabled everyone to express their opinion. As every coin has two sides, it paved a way for increased negativity and intolerance. In this context, unethical text includes offensive comments and hate speech targeting a group or an individual based on their characteristics such as race, religion or gender and that may threaten social peace. So, the unethical text should be eliminated before it reaches a user where identifying an immoral text plays a pivotal role. Various methods are used in identifying a text, one such method, widely known as Natural Language Processing (NLP) is used to perform text analysis in the process of detecting the unethical text. NLP enables us to perform tokenization, embedding matrix, padding and identification of semantic relationships, which helps to analyse the text data efficiently. Different algorithms were used to detect the unethical text, this study has used ISTM and Bi-LS TM to identify whether the user’s statement is offensive or not. The better and accurate model is considered and a web page is built on it. It predicts the offensiveness of text present in user input. In this way, the proposed model detects the unethical text on the web, which can be used to eliminate unfiltered comments. The best model is achieved by using the Bi-ISTM with an accuracy of 86.4%.