Automatic User Preferences Elicitation: A Data-Driven Approach

[Context and motivation] In the increasingly competitive software market, it is essential for software companies to have a comprehensive understanding of development progress and user preferences of their corresponding application domain. [Question/problem] However, given the huge number of existing software applications, it is impossible to gain such insights via manual inspection. [Principal ideas/results] In this paper, we present a research preview of automatic user preferences elicitation approach. Specifically, our approach first clusters software applications into different categories based on their descriptions, and then identifies features of each category. We then link such features to corresponding user reviews and automatically classify sentiments of each review In order to understand user preferences over such feature In addition, we have carefully planned evaluations that will be carried out to further polish our work. [Contributions] Our proposal aims to help software companies to identify features of applications in a particular domain, as well as user preferences with regard to those features. We argue such analysis is especially important for startup companies that have few knowledge about the domain.