Dr.T.VelumaniM.Sc(CS),M.Phil,MBA,B.Ed,M.Sc(Psychology),Ph.D is working as an  Assistant Professor in Department of Computer Science at Rathinam College of Arts and Science(Autonomous),Coimbatore 641 021 from 14-09-2020 to till date. He worked as Assistant Professor in Department of Computer Science Kongu Arts and Science College (Autonomous), Erode from 11-06-2009 to 30-03-2020 TamilNadu, India. He has got more than 11 years of teaching experience. He has obtained his B.Sc(CT),M.Sc(CS), M.Phil and MBA Degrees from Periyar University  Salem, TamilNadu, India. He has obtained his B.Ed Degree from Indira Gandhi National Open University (IGNOU) at Delhi, his M.Sc (Psychology) from Madras University Madras, TamilNadu, India. He has obtained his Ph.D in Computer Science from Manonmaniam Sundaranar University at Tirunelvelli, TamilNadu, India. His area of interest is Image Processing research directions involved knowledge discovery and data mining, pattern recognition, knowledge-based neural networks, Software Engineering and IOT.

An Intelligent Linguistic Attention Framework with Stacked Bidirectional Context Modeling for Multiclass Hate Speech Classification The extensive utilization of social media and other online platforms has cleared the way for unmatched communication and information exchange. Conventional methods for hate speech detection, to name a few being, matching of keywords, rule-based mechanisms and application of machine learning (ML) algorithms, frequently scuffle to obtain fine-drawn and context-dependent characteristic of hateful content. Implementing Artificial Intelligence (AI)-powered solutions can help analyze complex linguistic patterns and contextual information in detecting multi-class hate speech leading to substantial savings in training time and minimizing false positive rate.  This study proposes an AI-powered deep learning method called, Adaptive Moment-optimized Linguistic High-level Attention with Stacked Bidirectional Contextual Memory (AMLHA-SBCM) for multi-class hate speech detection, leveraging the malignant comment classification dataset. The AMLHA-SBCM method includes Global Complex Linguistic High-Level Attention Layer-based Keyword Extraction to avoid automatic filters depending on simple keyword blacklists and utilize subtle mechanisms to express hate, therefore minimizing false negative rate. Also Stacked Bidirectional Contextual Information Long Short Term Memory-based Keyword Extraction is used to derive surrounding elements for differentiating between hate speech from benign cases, therefore reducing false positive rate.  Also by stacking the Bidirectional Contextual Information Long Short Term Memory helps in learning progressively complex pattern that a single layer lack in obtaining dependency, therefore aiming in high precision and accuracy. Finally, by optimizing the learning rate using Adaptive Moment Estimation based optimal hyperparameter selection, the method aims to improve training time and enhance the overall efficiency of multi-class hate speech detection process. Notably, the AI-powered AMLHA-SBCM method exhibited superior performance, achieving an impressive accuracy of 33%, F1-score of 28% and training time of 38%, demonstrating evolutions in hate speech detection and contributing to a safer online environment.