Abstract Traditional customer segmentation methods predominately rely on high-cost market survey, in which data is subjective. As a real expression of customers, online customer reviews (OCR) contain valuable information for customer segmentation analysis. To solve OCR-based customer segmentation problem, this research proposes a two-stage customer segmentation framework. In the user modeling stage, an synonymous attribute recognition method and an attribute utility conversion method are designed, combining with tree structure of product attributes, to build a granularity unified user preference vector; in user clustering stage, Fuzzy C-Means (FCM)-based clustering process including an optimal cluster number identification method is developed for user clustering. Empirical studies justify the effectiveness of the proposed method.
|