Temporal association rules mimng(TARM) is widely applied in many applications, it aims at mining rules within a certain interval of time. Most of the exiting algorithms for TARM need to scan several times of the database, or do not consider the different exhibition period of an individual item, so the efficiency of these algorithms are not enough. In this paper, we present a novel approach to investigating TARM, the proposed algorithm works in an incremental way which takes the different exhibition period of individual item into account, in order to reduce the cost of storage, and only the frequent 1-I-tem is stored, and efficient pruning techniques are adopted to reduce the scan times of the database, and it only needs at most one time to scan the whole data set to obtain all the temporal association rules. The experimental results show that the proposed algorithm is efficient.%时态关联规则挖掘是针对在一段时间范围内的关联挖掘,在现实中有较多的应用.现有的大多数时态关联挖掘算法或者需要多次扫描数据库,或者没有考虑各个项在数据集上出现或结束时间上的不同,因而挖掘性能受到较大的制约.为此,本文提出一种增量式的面向具有不同时间出现与结束的项的时态关联规则挖掘算法.为减少存储方面的开销,只需保存已挖掘过的历史数据集中的频繁1项集.为了减少数据的扫描量,通过有效的剪枝策略,有选择性地扫描相关事务项,至多只需扫描一次完整的数据库.实验证明,该算法具有较好的挖掘性能.
展开▼
