Research On Improving Classic Algorithm Of Association Rules And It's Application In Bussiness Intelligence

As the development of the Database technology are becoming more mature and widespread,the amount of data which people collected is growing in exponential speed.When the amount of data grows extremely, we will feel lost in front of the information sea if we don't have an effective method ,with the help of computer to distill useful informations and knowledges. To coping with this challenge, Data Mining technology is given birth. Data Mining(DM) is definded as mining knowledges that people get interested from mass data. It is one kind of data analyzing method in deep level, which has been considered as an effective way to solve so called"Data explodes but knowledge lacks"problem. In recent years, it has been researching all over the database world. After practices and researches of several years, its economic value has shown up and has been largely applied in scientific researches, financial investments, marketing, insurance, medicine, production and management of communication networks.Highly effective mining arithmetic has always been put on an important place through the whole data mining research. As we face a mess of data sets, the efficiency of the arithmetic is the key. But what we have now have always drawbacks here and there, and because of that, this article puts more attention on mining arithmetic of associated rule.On the basis of having researching on Apriori which is a classic arithmetic, this article gives a new one, which improves Apriori from two aspects. First, it reduces the time consume; second, it can mine rules hiding behind small probability events. That is to say, on one hand, as the potential frequent item sets produced by Apriori have too big scales, the new arithmetic tries best to reduce the scale of the potential frequent 2-item set and to get even closer to frequent 2-item set. By doing this, the time consume greatly reduced. On the other hand, as the items in the database are not even, the probabilities are very different from each other, which results in getting rules that cannot concerning with items with low probabilities. So we gave them different weights and we can get very valuable rules that Apriori can't.