Through extenics-based case reasoning, this research puts forward a method to configure the modules of complex machine products. Firstly, the knowledge on the process of product design was represented, and the convergence modes of design plans were analyzed, with the aid of product semantics and the knowledge representation model of the primitives. Next, the semantic correlation function was constructed, and a priority degree evaluation method for product family was created, in order to obtain the optimal configuration plan. In this way, the excellent design genes in product plans with relatively low correlations will be preserved. Finally, the proposed method was verified on the design of computer numerically controlled (CNC) roll grinder. Specifically, the quantitative and qualitative user needs were converted into the design objectives, producing the set of primitives (design objectives). Based on sematic correlations and priority degrees, the product/product family that best match the conceptual product was searched for in the case library, and the optimal configuration plan was prepared, in the light of semantic correlation network and local correlation evaluation. The results show that the local correlations of the bed and workbench module, headstock module, grinding system, measuring system, and support of MK8420 with design objectives were 0.298, 0.450, 0.600, 0.500, and 0.298, respectively, indicating that the configuration plan is in line with user needs and suitable for case-based reasoning. The proposed method has been successfully applied in actual production, laying a good basis for the intelligent design of products.