.png){kind=icon}
.png){kind=icon}
Return
Supplier behavior modeling and winner determination using parallel MDP
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eswa.2010.08.044.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Behavioral uncertainty of a supplier is a major challenge to a buyer operating in e-procurement setting. Modeling suppliers' behavior from past transactions, estimation of possible future performance and integrating this knowledge with the winner determination process can bring a new dimension to procurement process automation. We propose a states-space model to capture the uncertainty involved in long-term supplier behavior. The states represent the performance level of a supplier. This behavioral aspect is then integrated with the winner determination process of a multi-attribute reverse auction for efficient supplier selection using parallel MDP. We also propose an implementation framework to collect the feedback on supplier, generate an aggregate performance score and integrate it with the winner determination process. The performance aggregation and winner determination with help of Markov decision process effectively uses the past performance information. In addition, it updates performance information in regular invervals and allevates the problem of maintaining a long history. We compare the MDP-based selection with that of performance score-based selection through a simulation experiment. It is observed that our scheme gives better buyer utility, selects best suppliers and fetches better quality product. The benefits realized through these attributes to the buyer increases the efficiency of the MDP-based selection process. (C) 2010 Elsevier Ltd. All rights reserved.
Keywords:
e -Procurement
Supplier behavior
Supplier performance
Markov decision process
Multi-attribute reverse auction
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W