Recommender Systems for Variant Management in the Automotive Industry

Abstract

This paper transfers some state-of-the-art methods of recommender systems for an application in the product development process of variant rich products in the automotive industry. Therefore, an introduction into the characteristics of the rule-based description of variant-rich products is given, followed by a presentation of three selected recommendation approaches, namely Collaborative Filtering, Association Rule Mining and Bayesian Networks. The presented approaches are then evaluated against the background of the variant-rich product configuration. Advantages and disadvantages of the methods in regard of this special use-case are highlighted and possible applications and limitations are discussed. In conclusion, further research needs for future implementation are identified.