A Methodology For Comparison Of User Reviews With Rating Of Android Apps Using Sentiment Analysis

On the daily basis,there are millions of applications,which are being uploaded by the developers.The millions of users download these applications without any check of duplicated applications.These applications hit the personal information of the users that damage the users' trust in Google store and also make the loss for users.Aggregately,there is a little information about these applications,which is given on a play store.Google play store allows the user to download an app and give him feedback in the form of rating and text reviews.A recent study analyzes that user requirement;the idea for improvement,user sentiment about specific features,and descriptions of experiences is beneficial for an application developer.However,many application reviews are extensive and complicated to process manually.Star rating is given of the whole application,and the developer cannot analyze the single feature.Star rating has been analyzed to not be authentic in comparison to user reviews because after knowing about the application,the users give reviews according to the personal experience.Because the main problem is that most of the applications that show the top ranking in the rating graph have been found to be incorrect in reviews sentiment analysis graphs.In this research,thousands of user reviews have been scrapped through different data scraping techniques.The text classification has been applied on these user reviews.Different machine learning techniques have been applied to check how many positive,negative,and natural reviews of an application also found a ranking of applications.The correlation between the average sentiment analysis and the average rating of the applications has been analyzed.