A Java tool to automatically recognize natural language fragments in user reviews that are relevant for developers to evolve their applications.
It processes and classifies reviews in natural language and generated aggregated summaries. These contain for example potential feature requests, reported issues or frequently asked questions. The latter could for example indicate that your app might not be fully intuitive to users.
It is to mention that my app powernAPP played an supporting role in the published whitepaper with the title ARdoc: App Reviews Development Oriented Classifier (PDF). The many reviews of powernAPP served as test data for the proposed ARdoc system. If you take a closer look at the published video, then you might notice that the shown reviews are from my app. Last but not least, they mentioned me in the acknowledgment section of the paper.
We thank Benjamin Sautermeister and André Meyer for helping us to evaluate the accuracy of ARdoc validating the results of the automatic classification on user reviews of their mobile apps. Sebastiano Panichella gratefully acknowledges the Swiss National Science foundation’s support for the project “Essentials” (SNF Project No. 200020−153129).
Personally, I think such a tool could be very helpful for (indie) app developers like me. In the meantime, I have 22 apps in the Windows Store. In total, I receive around 10 to 100 Reviews each single day. Therefore, without any sort of automation, it is almost impossible to read and respect each of them individually.