Information Recommendation Based on Domain Knowledge in App Descriptions for Improving the Quality of Requirements

Abstract

Apps have concentrated sale platforms, in which there often exist some products similar to the new App to be developed. The main features of these products are given in their introductions, providing an important resource for developers to improve the quality of the requirements of their own App. In this paper, we propose an approach to gain and recommend requirements related information from App descriptions to help developers use the resource efficiently. First, we construct a model by mining domain knowledge from App descriptions with the method proposed in our previous work and use initial requirements to retrieve their related information from the model. Then, we analyze the information and recommend them from three aspects: static information of the existing Apps for identifying the priorities of requirements; functional features and non-functional properties of features for giving the detailed design of the Apps in requirements; and the combinations of features for enriching the requirements. To validate the proposed approach, we conducted experiments and a survey based on the data in Google Play. The results show that our approach can identify the existing products with initial requirements reasonably, and also indicate that the developers confirm the usefulness of the recommended information in practice.