Personalized configuration rules extraction in product service systems by using Local Cluster Neural Network

Purpose – Configuration systems are used as a means for efficient design of customer tailored product service systems (PSS). In PSS configuration, mapping customer needs with optimal configuration of PSS components have become much more challenging, because more knowledge with personalization aspects has to be considered. However, the extant techniques are hard to be applied to acquire personalized configuration rules. The purpose of this paper is to extract the configuration rule knowledge in symbolism formulation from historical data. Design/methodology/approach – Customer characteristics (CCs) are defined and introduced into the construction of configuration rules. Personalized PSS configuration rules (PCRs) are thereby proposed to collect and represent more knowledge. An approach combining Local Cluster Neural Network and Rulex algorithm is proposed to extract rule knowledge from historical data. Findings – The personalized configuration rules with CCs are able to alleviate the burden of customers in expressing functional requirements. Furthermore, in the long-term relationship with a customer in PSS realization, PSS offerings can be reconfigured according to the changing CCs with the guide of PCRs. Originality/value – The contribution of this paper lies in introducing the attribute of CCs into the antecedents of PCRs and proposing the neural networks-based approach to extracting the rule knowledge from historical data.