Procurement auction WDP model based on SOM and DEA

There are some features existing in the procurement auction, for example huge number, multi-species, profit differences and irrational evaluation index systems. To solve these problems, this paper proposed the following model. Firstly, after vendors finishing biddings, we adopted self-organizing map (SOM) and bundle algorithm to simplify the vendors' biddings. This can enhance the goods' similarity within the bundles, balance the competition among suppliers, and reduce the complexity of the Winner Decision Problem (WDP) algorithm. Then, using the C2R model in Data Envelopment Analysis (DEA), we conducted a relative evaluation of simplified biddings and determined the final winner supplier. This can solve the problem of the uncorrelated property assumption and artificial weights in traditional evaluation index system method. Our model provides a new idea for WDP in the procurement auction. It has a good practicability.