Winner determination in online multi-attribute reverse auction under incomplete information settings

The 'winner determination problem(WDP)' is a computationally hard problem in multi-attribute reverse auction (MARA), especially when there are incomplete or indeterminate information about auctioneer's preference and the values of commodity attributes in sellers' bids. In the paper, a method for winner determination in online MARA was proposed, and an algorithm of the method was developed based on the theories of multi-criteria decision analysis (MCDA). All the bids received by auctioneer are evaluated through pair-wise comparisons and the preference degree between paired bids is calculated according to the conception of possibility degree. Finally, all the bids are ranked based on their ranking indexes and the winner of the auction is determined. It was demonstrated that the method can incorporate several types of incomplete or indeterminate information as long as they can be expressed in the form of linear inequalities. Moreover, it was proofed that the ranking order be consistent with the partial information on auctioneer's preference. A computational example was presented which verified the efficiency of the method.