Extracting and summarizing affective features and responses from online product descriptions and reviews: A Kansei text mining approach

Today's product design takes into account the affective aspects of products, such as aesthetics and comfort, as much as reliability and physical quality. Manufacturers need to understand the consumers' affective preferences and responses to product features in order to improve their products. Conventional approaches use manual methods, such as questionnaires and surveys, to discover product features and affective preferences, and then correlate their relationships. This is one-time, labour-intensive, and time-consuming process. There is a need to develop an automated and unsupervised method to efficiently identify the affective information. In particular, text mining is an automatic approach to extract useful information from text, while Kansei engineering studies product affective attributes. In this paper, we propose a Kansei text mining approach which incorporates text mining and Kansei engineering approaches to automatically extract and summarize product features and their corresponding affective responses based on online product descriptions and consumer reviews. Users can efficiently and timely review the affective aspects of the products. In order to evaluate the effectiveness of the proposed approach, experiments have been conducted on the basis of public data from Amazon.com. The results showed that the proposed approach can effectively identify the affective information in terms of feature-affective opinions. In addition, we have developed a prototype system that visualizes product features, affective attributes, affective keywords, and their relationships. The proposed approach not only helps consumers making purchase decisions, but also helps manufacturers understanding their products and competitors' products, which might provide insights into their product development.