摘要
通过观察可以发现连续近邻查询中KNN发生改变的必要条件是第个邻居发生变化,因此不需要监测所有近邻,只需要监测第个邻居即可。该方法采用边界线来监测第个邻居的变化,不过这需要将原始空间转变为时间-距离(TD)空间后进行操作。在TD空间中每一个对象用一个时间函数来表示,通过监测当前第个邻居的前视矩形区域来构造边界线。实验结果表明,边界线算法在非常大的时候是最有效的。
It is obvious that the necessary condition of changes in the KNN is the change of the k-th neighbor by observation, so the k- th nearest neighbor was monitored instead of monitoring all k （nearest） neighbors. A beach-line algorithm is developed to monitor the change of the k-th neighbor, but it needs transform original space into time-distance （TD） space for operating. In TD space, each object is represented by a function of time and the beach-line is constructed by following the rectangular looking-forward area of the current k-th neighbor. An experiment shows that the beach-line algorithm outperforms the most efficient when k is very large.
出处
《计算机工程与设计》
CSCD
北大核心
2008年第18期4759-4761,共3页
Computer Engineering and Design
关键词
空间数据库
移动对象
连续k近邻
边界线
时间-距离空间
spatial database
moving objects
continuous k nearest neighbor
beach-line
time-distance space
