Deep Learning Approaches for Sentiment Analysis: Comparative Insights into Sequential, Convolutional, and Transformer Architectures Vandana Jaiswal, Mahesh Parmar, Nishant Jain Proceedings 2025 IEEE Delcon International Conference on Recent Smart Technologies in Engineering for Sustainable Development, 2025 Sentiment analysis, a significant task within Natural Language Processing (NLP), plays a crucial role in understanding public opinion, customer satisfaction, and emerging trends through online platforms such as Twitter, Instagram, WhatsApp, and e-commerce portals. We proposed a hybrid sentiment analysis model (CNN+BERT) leveraging deep learning techniques to achieve superior accuracy compared to individual models. Using the Twitter US Airline Sentiment dataset, preprocessing steps including data cleaning, label encoding, train-test splitting, and tokenization were applied, followed by the implementation of CNN, LSTM, GRU, BERT, DistilBERT, and a proposed hybrid (CNN+BERT) model. Experimental evaluation revealed that the hybrid model achieved the highest accuracy of 92.46%. The proposed approach has potential applications in brand monitoring, product review analysis, market research, customer feedback evaluation, and real-time trend detection.
Novel Framework for Performance Prediction of Small and Medium Scale Enterprises: A Machine Learning Approach Nishant Jain, Abhinav Tomar, Prasanta K. Jana 2018 International Conference on Advances in Computing Communications and Informatics Icacci 2018, 2018 The small and medium scale enterprises (SMEs) are the prime factor for economic growth and job creation in developing countries. The literature shows that only a small number of SMEs are successful in achieving exceptional performance and sustainable growth. Therefore, it is paramount to determine the socioeconomic factors that hinder their growth. Incorporation of machine learning and statistical methods for solving business problems has gained substantial interest in recent years due to an exponential rise in consumer data. However, processing and interpreting this data to support business decision making is demanding, thereby leaving the scope for advancement. Therefore, in this paper, we design a novel performance framework with four modules, each having different functionality and contemplates machine learning methods. The fundamental objective is to predict the impact of strategic planning on SME's performance so that it can sustain in current competitive markets. For the sake of validating the framework, an experimental case study is conducted for a particular module, i.e., PMM Module. The prediction results for PMM module are compared in terms of RMSE concerning RNN, GBT, and RF methods.
