Sentiment Analysis of Unified Payments Interface (UPI) Payment App Review using Machine Learning Algorithm

The paper “Sentiment Analysis of UPI Application Reviews using Machine Learning Algorithm” uses natural language processing (NLP) and machine learning to analyse user evaluations of Unified Payments Interface (UPI) applications for sentiments. The objective is to conduct a comparative analysis of different algorithms to better understand the user reviews of UPI applications and provide useful data for these digital platforms' strategic enhancements. Preprocessing is done on datasets, which includes text transformation with techniques like Word2Vec and Bag of Words. These techniques transform raw text into numerical vectors that are used by machine learning algorithms. Reviews are categorised into positive and negative sentiments using sentiment analysis algorithms. For sentiment classification, algorithms like C 4.5 (J48), Naive Bayes, and LSTM are used. Their effectiveness is assessed through comparative analysis. 90.02% was the highest accuracy recorded for sentiment classification of user reviews on Paytm. LSTM in conjunction with the word2vec skip-gram training model resulted in the attainment of this prediction. Results provide perceptions into user sentiments, facilitating purposeful enhancements in UPI applications that boost user contentment and adhere to changing digital payment requirements.