MetadataShow full item record
In this thesis, we analyze the sequential bargaining problem from a different perspective. Instead of taking a game theoretic approach, we model bargaining as a search problem and use a genetic algorithm to find an equilibrium outcome. Users, as buyers or sellers, only have to specify the product details and reservation price. The bargaining process is done by buyer and seller software agents. We have also developed a fully functional website Cyber Haggler (http://www.cyberhaggler.com) to illustrate our concepts. The software agents are based on realistic assumptions of bounded rationality and they learn from trial-and-error over time. We also compare Cyber Haggler to both Kasbah and EBay.com. Results show that our model can be easily implemented commercially on the Internet. Thus, we have been successful in modeling real world human-like bargaining on the Internet.