Information Extraction for Mobile Application User Review

Abstract

The growth of mobile e-commerce and the popularity of smartphones makes the intensity of mobile app users increase exponentially. The users can provide reviews related to their experience while using the application, this review can contain valuable information such as complaints or suggestions that can be used for further in-depth analysis based on reviews given. However, the large number of reviews makes it difficult to find and understand the information contained in each review. To solve these problems, this study proposes a model that can extract information in the reviews by categorizing and analyzing sentiments in each review using the text mining approach and machine learning techniques, we use several algorithms for sentiment analysis, classification and modeling topics that are popularly used by previous researchers. The output of this model is a collection of the most trending reviews that have been identified and classified as polarity sentiments and review categories. We had conducted a series of experiments to find the best model, the average sentiment precision of reviews is 85% and the best algorithm for classifying the reviews obtained using SVM with an average FI score of 84.38% using the unigram feature while the NMF works better compared to LDA in modeling topic reviews.