| Summary: | 碩士 === 東吳大學 === 資訊科學系 === 95 === The transactional database in daily operational systems contains abundant unknown knowledge. For example, from the data mining of the clients’ transaction details, the habit of consumption of the consumers can be understood. From the consumption history of the con-sumers, the relationship between the consumed products can be analyzed. According to this information, the related products can be combined for cross selling in order to provide the best customer services to customers and improve the marketing performance. Performance is very important to knowledge mining in the large-scale database. It is because there is large amount of information in the commercial database. In the algorithm of data mining, how we can seek the information we want from the large-scale database becomes a very important question. This study has integrated databases and SQL to implement FP-tree establishment of Generalized Association Rules in data mining algorithms. SQL is adopted to simplify the complexity of data establishment and mining process of the FP-tree. According to the experi-mental result, it shows that our proposed method spends less time of establishment comparing with the traditional FP-tree algorithm. It is a mining method of high performance Generalized Association Rules, which improves the implementation efficiency of system in order to raise the actual usage.
|
|---|
