The app market is extremely competitive, with users typically having several alternative app possibilities. To attract and retain users, it is imperative for developers to consider the ratings and reviews their apps receive. App reviews frequently contain feature requests, sometimes hidden among complaints. Developers use these complaints and requests to improve their apps, thus increasing their rating which is incredibly important for attracting new users. Unfortunately, developers of new apps are at a severe disadvantage: They do not have the benefit of existing reviews, with only the reviews of similar apps to potentially rely upon. To address this problem, we conducted a study and developed a novel technique that extracts feature requests from similar, existing apps to help prioritize the features and requirements important in an initial app release.
We compared different classification models in order to identify most appropriate classifier for classifying reviews category-wise. We found that there is not one single classifier that could have a higher accuracy than others for all categories.Our study also involved extracting features from user reviews in the Google Play store. The features were presented to 17 Android developers twice; once without applying our technique and once after applying our technique. Our proposed technique created a 48\% reduction in the number of features considered high priority by participants; helping developers focus on what features to consider for their apps. We surprisingly found that the frequency of requested features did not impact the developer's decisions in prioritizing the features in the inclusion of new apps.
Library of Congress Subject Headings
Mobile apps--Development; Natural language processing (Computer science); Data mining; Text processing (Computer science)
Software Engineering (MS)
Department, Program, or Center
Software Engineering (GCCIS)
Saleh, Rebaz Saber, "How Should You plan Your App’s Features? Selecting and Prioritizing A Mobile App’s Initial Features Based on User Reviews" (2017). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus