Research on post-bidding bundle strategy for online reverse auction

Currently online reverse auction (e-RA) is often used in procurement. However, winner determination problem (WDP) is NP-hard problem and practical use is difficult. As a research focus, `bundling´ is an effective way to cut down the complexity of WDP for e-RA, and often applied in practice. Although there are some qualitative researches in this field, quantitative bundling is still a problem. This paper designs post-bidding bundle strategy (PBBS) for e-RA to focus on this issue. Based on mathematical model and the bidding data from suppliers, this strategy uses `item similarity´ and `market competitiveness´ to measure the possibility of two items in the same bundle. Specially, item similarity measure refers to two classic methods in scientometrics: Salton and Jaccard. Finally, as shown with the 7^th pharmaceutical centralized purchasing bidding data in china, this strategy of bundling has good theoretical value and practical significance in quantitative bundling.