Sentiment Analysis on Google Play Store Data using NLP

Abstract

Every day, thousands of users review mobile apps on the Google Play Store. User contentment, app ratings, and feature expectations are just a few of the important details included in these reviews. However, manually analysing this vast amount of unstructured material is challenging and time-consuming. The project suggests a sentiment-analysis system that uses Natural Language Processing (NLP) to automatically determine if user evaluations are favourable, negative, or neutral in order to address this problem. The procedure begins with gathering app reviews from publicly available datasets, after which the data is cleaned by eliminating special symbols, stop words, emojis, and undesired characters. NLP methods including tokenization, stemming, and word embeddings are used to process the cleaned text. A machine-learning model such as Random Forest, Decision tree or multilayer perceptron models are trained to classify the sentiment of each review.