Abstract
To accommodate the diverse users demands for consumer products, enterprises need to design and develop different lines of products according to different groups of users. Dynamic internet data, including product reviews, user attributes, and product configurations, are utilised to model users' stochastic product choice behaviours and mine the product design requirements of features, performance levels, and quantity. First, the web crawler is applied to collect internet data, and then the data are structured and the demand information is retrieved. Second, a product choice model is employed to capture the heterogeneity and correlation of user demands on product features. In particular, users' implicit requirements in terms of product function and performance are elicited from the text mining of product reviews. Third, incorporating various user requirements mined from dynamic internet data, graph theory analysis is introduced into design generation, product improvement, and market analysis. A case study on Chinese smartphones is presented, where the results show that the proposed method is practical and suitable for product-design analysis using the large volume of dynamic internet data.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 The new features of internet data refer to point (1) discussed in section 1.1: as the number of online products is large and the existence of social network and influence, the IID assumptions of products or consumers are not suitable.
2 JD is the second biggest online retailer in China, following Alibaba.
3 ZOL is the biggest technical forum in China for electronic devices.
4 For more information: https://code.google.com/archive/p/word2vec/; https://www.tensorflow.org/tutorials/word2vec
5 For ease of presentation, the lower case of time is ignored.
6 We thank Associate Professor Chaolin Yang from the Shanghai University of Finance and Economics to point out this interesting point and the possible explanation for the phenomenon.
7 According to reports from IC insights, Nielsen and Talking data, Vivo is a large Chinese smartphone company. However, we have much fewer reviews from JD for this brand than other brands. Therefore we did not consider Vivo in out case study.
8 α should be specified differently for each product i, because only the difference in utility matters in the logit model. If α is specified as the same for each product, α is effectively not used.
9 Based on a survey for approximately 30 colleagues and students to find out how many reviews they would read when buying a smartphone online. New specification should be considered while different datasets are used.