| Summary: | 碩士 === 南台科技大學 === 資訊管理系 === 95 === As the information technology is developed rapidly, enterprises have more and more channels to obtain information, the data are relatively more and more complicated too. These data include a lot of hiding information, how to excavate hidden information out, the method is called data mining. Excavate the rule from the course of prospecting, can take as the reference information while making policy in enterprises. Generally probing into the association rule is more referring to mining positive association rules, these rules can point out the positive association between data items. For example, the coustomer will buy the milk when buying the bread. As to most application, this positive information is useful. But as to some situations, the information of providing is still not enough, it is unable to make the entire aspect the consideration. Therefore, still need other useful information to help decision to go on, so account for the concept of negative association rules.
In this thesis, we will take FP-tree algorithms as the base foundation mining positive and negative association rules. And we will modify FP-tree in order to build a new data structure that just need scan database one time to mining frequent itemsets, let it more efficient in mining positive and negative association rules. Finally, the algorithm is applied to the dynamic databasse. When the data variation is not scan database again, the algorithm can be aimed at variation data mining, find out the intact and efficient positive and negative association rules.
|
|---|
