A Data Mining Approach for Analyzing Dynamic User Needs on UGC Platform

Nowadays, the official platform for consumer community has become a reliable database for enterprises to mine users' needs. The study aims to develop a dynamic demand mining method based on users' online reviews. To achieve this objective, this research proposes i) a data crawling process of online product reviews; ii) multi-dimensional index system of data processing; iii) dynamic user demand mining and transformation method. Particularly, complete user comments data within about 5 mouths on Pollen Club (Huawei official consumer community) were collected. After the initial data cleaning, a primary study of 2100 selected pieces of data was put forward. The methods for Chinese natural language processing (e.g., text segmentation, sentiment analysis) were integrated to process the data, and frequency analysis, trend analysis, cluster analysis and user analysis were used to mine the dynamic user data. The improvement suggestions on product features were put forward based on the theory of design knowledge hierarchy (DKH). This research is expected to help enterprises to mine dynamic user data more efficiently, discover the consumer feedback on the product performance, and facilitate further product improvement.