【作者】 饶丹；

【导师】 卢炎生；

【作者基本信息】 华中科技大学 ， 计算机软件与理论， 2004， 硕士

【摘要】 数据挖掘作为数据库研究领域中的热点,正受到越来越多的关注。它被定义为在数据中寻找正确的、有趣的、潜在有用的并最终可以理解的模式。基于关联规则的挖掘是其中一个重要的研究方法,具有重要的理论价值和广泛的应用前景。关联规则中最有名和最有影响力的算法是Apriori算法,Apriori对布尔型数据库的标准关联规则的挖掘有很高的效率。在很多领域中,只挖掘出标准关联规则是不够的,需要对数据项的否定项进行挖掘。带否定项的关联规则是指允许在关联规则中出现负项目,对关联规则的形式做出了扩展,从而提高了关联规则的描述能力。 Apriori算法在挖掘带否定项的关联规则时显得非常笨拙和低效,不擅长处理相关性高的数据,CA算法对稠密型数据库有很好的效率,但是挖掘出来的规则会很多,并且很多规则可能是无趣的。针对这种不足,INA算法在支持度-置信度的模式上增加了一个新的评估参数-兴趣度。通过引入兴趣度,有效的改善了CA算法中的候选集膨胀现象,删除了一些被挖掘出来而没有兴趣的关联规则。这种算法能够有效挖掘出有趣的带否定关联规则。INA算法在多维数据库中挖掘带否定关联规则也有较好的效率,更广泛的应用了带否定关联规则的挖掘。对于多维数据库,采用与数据仓库相结合的方法,利用数据仓库联机分析的一些结果,能够更好的挖掘带否定关联规则。通过初步实验将CA算法挖掘出的频繁项目数与Apriori算法挖掘出的正频繁项目数进行了比较,显示了CA算法的有效性,同样将INA算法挖掘出的有趣规则与CA算法挖掘出的规则进行比较,表明了INA算法的实用性。更多还原

【Abstract】 Data Ming is recognized as the hot topic in the database research field, and has received increasing attention by researchers. Which is defined as the non-trivial process of identifying valid, novel, potentially useful and ultimately understandable patterns in data. Mining association rules from large databases plays an essential role in many data mining tasks and has broad applications.Association rules mining is considered to be one of the most important tasks of Data Mining, which is to find the rules with the support and confidence greater than the thresholds. The most famous and effective algorithm in association rules mining is Apriroi algorithm. The algorithm is effective at standard association rules mining in Boolean database.In many fields, only mining standard association rules are not enough. Mining negative association rules is required. Negative association rules allow the negative itemsets appear in the association rules, expand the defining of association rules, and then increase the ability of description of association rules.Because between the negative itemsets and the positive itemsets has high relativity, Apriori algorithm is clumsy and lowly effective in mining negative association rules, and it is not good at deal with the high relative data. Based on the relativity, CA algorithm has good efficiency in close database. But it can mine many association rules including useless ones by this way. INA algorithm adds a new evaluating parameter into the support-confidence model –interesting parameter, which improves the performance of candidate expansion in CA algorithm, and deletes many uninteresting association rules. INA algorithm is effective in the multi-dimension, which apply the negative association rules in more fields. To multi-dimension database, combining the data warehouse is more effective in mining association rules.Comparing the frequent itemsets number with CA algorithm and the positive frequent itemsets by experiment, the experiment shows the validity of CA algorithm. <WP=5>In the same way, comparing the interesting association by CA algorithm with by INA algorithm, and the experiment shows the practicability of INA.更多还原