Case Based Reasoning in E-Commerce.

  • Zhaohao Sun

Student thesis: Doctoral Thesis


The revolution of the Internet and the WWW are changing traditional commercial activities such as shopping, brokerage, negotiation and retailing. Intelligent techniques for e-commerce have drawn increasing attention since the end of the last century. However, how to make e-commerce intelligent and the customers more satisfied remains a big issue. Applying intelligent agents or multiagent systems and CBR in e-commerce has been among the most rapidly growing areas of research and development in information technology in the last few years. CBR potentially has a large role to play in facilitating e-commerce, because it is experience-based reasoning, which plays an important role in business. However, applying CBR in multiagent e-commerce is still in its infancy, although there are some studies on CBR in multiagent negotiation and auction. There are also no systematic studies on integration of CBR, MAS and e-commerce from both a mathematical and a logical as well as an information technology viewpoint. There are few studies on applying CBR in multiagent brokerage.

This thesis will fill this gap by examining intelligent techniques such as case-based reasoning (CBR) and their applications in e-commerce, and providing a unified treatment of integrating CBR, MAS and e-commerce.

The philosophy of the thesis is that just as human agents play an important role in traditional commerce using their intelligence, intelligent agents will play the same role in e-commerce through their possessing intelligent techniques. In order to realize this philosophy, this thesis will make some contributions to CBR, intelligent agents and MAS, and e-commerce. Three of them will be briefly mentioned as follows, while the rest will be mentioned in the concluding remarks of each chapter in this thesis. The first contribution of this thesis is to provide a general theory of CBR based on similarity-based reasoning, in which the thesis introduces a new theory of similarity metrics, a novel process model for CBR (the R5 model), examines abductive CBR and deductive CBR, and develops algorithms of rule-based and fuzzy rule-based case retrieval. It also shows that CBR is a process reasoning, in which a traditional reasoning paradigm plays a pivotal role in each stage of the process.

The second contribution is to develop efficient intelligent techniques and methodologies for multiagent e-commerce, in which the thesis provides deeper insight into multiagent e-commerce by classifying it into three categories: multiagent auction systems, multiagent mediation systems, and multiagent brokerage systems.

The final contribution of this thesis is to develop the unified methods, models and architectures for multiagent e-commerce, in particular for multiagent brokerage, and then integrate CBR, multiagent systems, and e-commerce in CMB, which is a CBR system for multiagent brokerage.

In order to make the above mentioned contributions, the thesis is undertaken at three different levels: a theoretical level, a technological level, and an implementation level.
Date of Award4 Oct 2003
Original languageEnglish
SupervisorGavin Finnie (Supervisor)

Cite this