Feature model augmentation with sentiment analysis for product line planning

A feature model is able to identify commonality and variability within a product line, helping stakeholders configure product variants and seize opportunities for reuse. However, no direct customer preference information is incorporated in the feature model when it comes to the question-how many product variants are needed in order to satisfy individual customer needs. This paper proposes to mine customer preference information for individual product features by sentiment analysis of online product reviews. The features commented by the users of a product are used to augment a simple feature model predefined with customer opinionated preference information. In such a way, the customer preference information is considered as one attribute of the features in the model, helping designers make informed decisions when trading off between commonality and variability of a product line. Finally, we present a Kindle Fire tablet case study to demonstrate the proposed method.