Research On Recommendation System For Customized Products Based On Product Platform

With the market developing in the direction of personalized customization, mass customization becomes the inevitable trend of manufacturing industry and the main mode of production in the future for its low cost, diversification of products and rapid response to market. In the context of mass customization, many companies use the product platform technology, which establish a common set of modules and components. On this basis, a series of product variants can be quickly developed to meet diverse customer needs.Currently, most of the customized products based on product platform are recommended to customers by human experts. A few systems of customized products guide customers to configure products on-line by tree-like or wizard-like methods. By using these systems, customers choose the components of product step by step from the list of options, and ultimately obtain customized products. For example, the Dell TM sells configured personal computer on the web site by using a wizard-style B2C application software(www.dell.com). However, these methods bring about a problem that configured part of products may have unbalanced performance, since customers may not be familiar with a product’s technical parameters and specific meaning.With the popularity of the Internet and development of E-Commerce, shopping online is changing the traditional business model. E-Commerce site has been a palce of business for an enterprise. And various forms of recommendation systems were used in these sites at different levels, such as Amazon, TaoBao and DangDang. Recommendation systems were proposed to suggest products and provide customers with information of which products to purchase, Recommendation systems can enhance sales by converting browsers into buyers, cross-selling and retaining customers effectively and so on. Successful E-Commerce recommendation system can produce huge economic benefits.In this research, by considering the characteristics of product platform, advanced E-Commerce recommendation algorithms are applied to recommendation system for customized products based on product platform. The customer’s consume history, preference and other information are used to help customers conduct a reasonable choice of customized products. It will meet the needs of customers and enterprises at different levels by its successful application. This research was supported by the National Natural Science Foundation (70871020) and basic research projects of Ministry of Education (N090404019). The research in the paper includes the following aspects:(1) Association rules is used to analysis the relations between the modules, and recommendations to user are generated by a recommendation algorithm of association rules.(2) Conjoint analysis is adopted to calculate the utility of the product modules, and recommendations are generated according to the the utilities calculated.(3) Collaborative filtering recommendation algorithm is used to solve the ratings sparsity problem generated in the process of user ratings. Customer preference can be predicted with items a user does not score and recommendations are generated.(4) A case of Cell phone is designed and developed as a recommendation system for customized products based on product platform. The proposed recommendation algorithms are embedded into the system, and the corresponding recommendation functions are implemented.