摘要
传统的联结规则挖掘算法依赖于一个不现实的假设:用户可以指定最小支持度。如果用户不了解他们的数据库,指定的最小支持度是肯定不适合的。在此设计了一个基于遗传算法的挖掘策略,它具有两个显然的优点:①高性能且自动化的规则挖掘;②不要求用户指定最小支持度。
The performance of Apriori-like algorithms for identifying frequent itemsets relies upon the user-specified threshold of minimum support. If a minimum-support value is too large,very few frequent itemsets might be found in a database. In contrast,a slightly small one might lead to low-performance (too many frequent itemsets). This generates a crucial challenge:users have to give a suitable minimum-support for a mining task. However,it is impossible for users to provide such a suitable minimum-support if they have no knowledge concerning their databases. This paper designs an evolutionary strategy for mining generalized association rules without minimum support. Using such strategy, two benefits are delivered: (1)the approach is effective and efficient for global searching, especially when the searching space is so large that it is hardly possible to use deterministic searching method; (2)system automation is implemented because this model does not require the user-specified threshold of minimum support.
出处
《广西师范大学学报(自然科学版)》
CAS
2003年第4期22-31,共10页
Journal of Guangxi Normal University:Natural Science Edition
基金
Large Grant of the Australian Research Council(ARCDP0343109)
Grant of the UTS