Scopus Indexed Publications

Paper Details


Title
Deep Learning-based Sentiment Analysis of User Generated Reviews of Various AI Powered Mobile Applications
Author
Saymon Ahammad, Mahjabeen Hossain, Mustak Ahmed, Nur-A-All Asif, Nurul Afsar Ikram, Sadia Akter Sinthia,
Email
Abstract

AI powered apps offer numerous benefits and positive impacts on various aspects of daily life, offering innovative solutions, personalized experiences, and enhanced efficiency across diverse domains. Using these AI powered apps (ChatGPT, Imagine, Replica, Bing, Alexa, Lensa, Arta, Ask AI, Answer. AI, and Write Cream AI.) are very easy, as these apps are commonly available in the Google Play Store. This research delves into the analysis of sentiment within user reviews of AI Google Play Store apps. By using 122231 data points that have been classified as neutral, negative, and positive feelings, our aim is to apply advanced deep learning algorithms to determine patterns in sentiment expression. Through meticulous data preprocessing, the unstructured dataset is transformed into a well-structured format, enabling insightful analysis. Employing six distinct deep learning methods on the categorized dataset, including CNN (90.53%), LSTM (91.97%), BiLSTM (91.67%), GRU (91.44%), BiGRU (91.34%), and CNN+LSTM (92.1%), this study has achieved notable accuracy rates. Particularly, the pre-trained CNN+LSTM model emerges as the best performer, attaining the highest accuracy of 92.1%.

Keywords
"Artificial Intelligence , ChatGPT , Bing , Google Play Store , Deep Learning , Convolutional Neural Network , Natural Language Processing , Sentiment Analysis"
Journal or Conference Name
7th International Conference on Inventive Computation Technologies, ICICT 2024
Publication Year
2024
Indexing
scopus