Analysis of Economic Impact of Online Reviews: An Approach for Market-Driven Requirements Evolution

As a novel market data, online reviews can manifest user demands in real contexts of use. Thereby, this paper proposes a demand-centered approach for requirements evolution by mining and analyzing online reviews. In our approach, it is challenging to improve the accuracy of opinion mining techniques for huge volume of noisy review data. Furthermore, how to quantitatively evaluate the economic impact of user opinions for determining candidate requirements changes is also a challenging problem. In this paper, an opinion mining method augmented with noise pruning techniques is presented to automatically extract user opinions. After automatic synthesizing the information extracted, a utility-oriented econometric model is employed to find causal influences between the system aspects frequently mentioned in user opinions and common user demands for revising current requirements. A case study shows that the presented method of opinion mining achieves good precision and recall even if there is a large amount of noisy review data. The case study also validates the effectiveness of our approach that it discovers the candidate requirements changes related to the software revenue, especially the ones that are ignored by software developers.