Exploring Content-Based And Comment Classication Using Deep Adaptive Learning (KEY IJP************764)

• Anirban Sasmal,Manish Kumar

Abstract

In recent years, the prevalence of SMS as an primary communication channel has soared, making it an integral part of daily interactions. However, this surge in usage has also led to the emergence of SMS spam, a nuisance that not only disrupts users but also poses potential risks such as credential theft and data compromise. To combat this issue effectively, Natural Language Processing (NLP) techniques coupled with Deep Learning models have emerged as a promising solution, particularly for text classification tasks. Among these models, Long Short-Term Memory (LSTM) networks have demonstrated remarkable performance in binary and multi-label text classification problems. this paper presents an innovative approach that combines two distinct data sources: one targeting Spam detection in social media posts and the other focused on Fraud classification in emails. By merging these datasets and leveraging common bigrams extracted from each, we devised a multi-label LSTM model tailored for identifying malicious text across varied sources. Our experimental results underscore the effectiveness of this approach, showcasing the model's capability to discern malicious content irrespective of its origin. the LSTM model trained on the amalgamated dataset exhibited superior performance compared to models trained independently on each dataset. This enhancement in performance can be attributed to the model's ability to learn and generalize from a diverse range of text samples, thereby improving its predictive accuracy and robustness. Additionally, the utilization of common bigrams extracted from both datasets facilitated the model's understanding of recurring patterns and linguistic nuances associated with spam and fraudulent content. overall, our findings highlight the efficacy of employing LSTM-based Deep Learning models for combating SMS spam and related security threats. The approach presented in this paper offers a practical and efficient means of classifying malicious text, contributing to the ongoing efforts in ensuring a secure and trustworthy communication environment for SMS users.