CONTEXT EXTRACTION AND EXPRESSION OF PRODUCT IMPROVEMENT FEATURES BASED ON ONLINE REVIEW MINING
Keqin Wang, Li Deng, ANGQI LEI, Zhihong Huang, Qingyu Ma, CHEN ZHENG, Jinhua Xiao, Jing Liu · DYNA · 2025
Current research on extracting customer requirements from online reviews is mainly focused on feature extraction and sentiment analysis, often neglecting the context information of product use. This oversight limits the designers' ability to fully understand user interactions with products, hindering effective product design improvements. To address this issue, we propose a method for context extraction and expression of product improvement features based on online review mining. Firstly, a context model was constructed by identifying context elements related to product improvement features. Secondly, a conditional random field (CRF) model was trained to automatically annotate these elements in reviews. Lastly, Chi-square test was conducted to quantify correlations between context elements and product improvement features, ultimately creating a matrix diagram for visualizing design directions. The effectiveness of the method was validated using camera review data as a case study. Key Words: online reviews; product improvement features; customer requirements; context extraction; context expression