CONCLUSION

This chapter suggests a hybrid recommendation mechanism and a prototype system called CAR. The hybrid recommendation mechanism is based on association rule mining and on CBR, which is aimed at enriching the recommended information.

The proposed mechanism consists of a four phase-association rule generation, case generation, construction of a hybrid knowledge base, and hybrid recommendation. The result of our experiment with an illustrative web log database proved to be valid and robust.

In conclusion, this study shows how the tacit knowledge within a web site can be brought together to create valuable decision support tools for an Internet Business focused on B2C (Business to Consumer). It is expected that the proposed recommendation mechanism will have a significant impact on the research domain related to B2C Internet Business and CRM. Further research topics still remaining are as follows :

1. The basic technology of data mining used for this study needs to be improved so that more complicated customer knowledge can be analyzed .

2. CBR needs to be integrated with other artificial intelligence-based reasoning algorithms, such as fuzzy cognitive map (FCM), so that more complicated web-based decision problems can be analyzed effectively.

3. CAR, our prototype system, needs to be updated with other commercial functions so that more practical recommendation problems can be solved easily.