摘要
在当今类似车载系统中常有需要我们在物体时空运动数据库中进行数据挖掘,然后根据得到的规则预测物体的运动趋势的需求。至今,对于由时空两方面描述的数据进行挖掘的研究还没有特别明确的方法。提出MINE_ALLFP算法来发现所有的频繁集。为了增加研究的可行性,去掉位置信息的连续性,提出将整个大空间分割为若干个小区域的方法,这里还引进一种好的剪枝算法。
Explanations of movements of vehicles moving in traffic require descriptions of the patterns they exhibit over space and time. The field of spatiotemporal data mining where the data relationship is defined by the spatial and temporal aspects of data is still in its infancy. In this paper,weintreduce an algorithm,called AllMOP,to mine all frequent movement patterns in traffic data. Due to the imprecision of the sampled pasitions,they are represented by the regions whose sizes depend on the degree of the points＇ accuracy. In order to control the dense of the pattern region we apply agrid-based clustering method to the generation process. Moreover,with a good candidate pruning method the execution is reduced. Our technique outperforms the gridbased GSP technique with respect to data compression and memory. It is applicable to traffic monitoring,trafflc management,as well as traffic location-based service.
出处
《黑龙江科技信息》
2010年第5期56-56,55,共2页
Heilongjiang Science and Technology Information
关键词
运动模式
时空挖掘
位置预测
movement pattern
Spationtemporal data mining
location prediction
