Research on the Method of Constructing Product Innovation Demands Graph Based on Competitive Product Reviews

Nowadays, user review texts are the main source of product innovation demand for enterprises, but in the face of massive and unstructured user review texts, how to efficiently identify and visually express product innovation demand has become an urgent problem for enterprises. Although a series of studies have explored this issue, there is a lack of studies that cover the complete path of identifying, classifying, and visualising product innovation demand, and few studies have introduced reviews of competitive products for comparative analysis. Therefore, under the guidance of cognitive flexibility theory, this study firstly crawls the text data of user reviews of Huawei mate50 and Xiaomi 13, and then identifies the product innovation needs using BERT Fine-Tuning after data preprocessing; then the product innovation needs are clustered by themes using BERTopic, which classifies the themes into common and specific themes; finally, based on the theme clustering results, we use dependency parsing and association rules to construct a product innovation demand (comparison) graph under unique (shared) themes. The results of this research not only provide a new way of organising knowledge in the field of ill-structured knowledge, which is suitable for the development of cognitive flexibility theory in the era of big data, but also helps to guide enterprises to produce innovative products that really meet the needs of end-users, and promotes the transmutation of innovation from the production paradigm to the service paradigm in the era of knowledge economy.