Introduction

Consumers' online information searching and sharing behaviors leave traces, which can be analyzed to infer their interests and perception of the market. In the past few years, academics and practitioners have begun to realize the potential of online consumer forums, blogs, and reviews data. Researchers showed that what consumers are searching for provide indicators of their interest in a product, and are predictive of what they will shortly purchase (Kulkarni, Kannan, & Moe, 2012). Based on the correlation between online searching activities and purchases, several prediction models have been proposed to forecast product sales (Fantazzini & Toktamysova, 2015). To collect online data more efficiently, a handful of studies recently applied text-mining (Archak, Ghose, & Ipeirotis, 2007; Eliashberg, Hui, & Zhang, 2007; Ghose, Ipeirotis, & Li, 2012; Tirunillai & Tellis, 2012). Lee andBradlow (2011) text-mined semi-structured product reviews to understand market structure based on the similarity of the products' mentions with various attributes.