In the era of Web3.0, User Generated Content (UGC) has broken the One-to-Many Model of traditional media and become a new driving force and trend of media development. UGC types of text, picture and video have more and more influence on consumers’ products image perception. With the application of UGC in hotel industry research, it is a good attempt to study the consumers’ image perception of B&B (Bed and Breakfast) based on UGC data. In order to help researchers fully perceive the image of B&B, this research intends to talk about how to collect and sort out three types UGC data about B&B through text mining technology. Based on the theory of cognitive-affective model, SnowNLP and LDA model are used to analysis emotional image and cognitive image, which may help researchers to analysis the overall consumers’ image perception of B&B’s. Meanwhile, the paper will also talk about how to apply the received image perception of B&Bs to better provide B&B service.