Extracting Attractive App Aspects from App Reviews using Clustering Techniques based on Kano Model

Kano model is a technique to evaluate product features based on users’ satisfaction. The problem with the traditional Kano approach is that it is limited to the small amount of data collected manually from users and the sample of participants might not be representative. Many users are interested in evaluating product features via social media platforms. The limitation of the Kano model can be mitigated by using users’ feedback from social media as a source to understand their satisfaction. Nowadays, several mobile applications are developed to solve the same problem and serve the same domain. Hence, it has become difficult to compete with similar products and increase users’ satisfaction to win the market advantage. In this research, app reviews were analyzed using natural language processing and clustering techniques to extract app aspects that increase user satisfaction by labeling them according to the Kano model categories. The clustering was based on aspects dissatisfaction and satisfaction values. We evaluated the results of the clustering technique based on a ground truth that was built from a user survey. Experiments showed that our clustering and labeling approach was able to identify the attractive app aspects better than must-be and one-dimensional aspects.