关键词: 产品在线评论, 感性工学, 词频, 评估, 强度, 活动, BP神经网络, 产品设计

Abstract: Aiming at the problems of high subjectivity,low real-time and few data in traditional Kansei engineering,a Kansei engineering driven by online reviews of product data.By taking the smartphone information of an e-commerce platform as the data source,the TF-EPA method integrated Term Frequency (TF) with Evaluation-Potency-Activity (EPA) was proposed,which could extract Kansei words.To obtain a more reasonable emotional evaluation,the sensory evaluation was calculated through word clustering and adverb-scoring of online reviews.From the perspective of cognitive psychology,a BP neural network was used to construct the nonlinear mapping model between product parameters and user Kansei images to simulate the user psychological assessment mechanism.Experiment on the smartphone case study demonstrated the generalization capability and effectiveness of the proposed model.

Key words: online product reviews, Kansei engineering, term frequency, evaluation, potency, activity, BP neural networks, product design