Abstract: The moving objects in the field of urban computing tend to have more information except spatial and temporal information. ?And the current spatio-temporal index mostly only has the spatial and temporal dimension, and the two dimensions cannot manage the moving objects with more information effectively. Based on this problem, this paper proposes the ATPR-tree, a spatio-temporal index with attribute value, the index is transformed from the TPR-tree. In the first part, based on the CBR in the node of TPR-tree, it introduces the concept of Rating Interval（RI）. Then on the basis of RI, it changes the TPR-tree node structure and cost function. Finally, the insertion and deletion algorithms make changes correspondingly. ?In the experiments, the moving objects are randomly generated by GSTD, and it compares the ATPR-tree which takes attribute as one dimension of the index with another TPR-tree which not do so. The experimental results show that, for the queries with attribute value, the ATPR-tree which takes attribute as one dimension of the index can effectively reduce the number of node access and improve the efficiency of query.

Key words: spatio-temporal index, urban computing, attribute value, moving object

摘要： 城市计算领域里所处理的移动对象往往带有除时间、空间信息外更多的信息，而目前的移动对象索引大多只有时间、空间这两个维度，不能有效地对附带更多信息的移动对象进行管理。基于这一问题，提出了一种带有属性维度的时空索引ATPR-tree，这种索引由TPR-tree改进而来。在TPR-tree节点CBR的基础之上新加入了属性值区间（RI）的概念；根据加入的RI属性维改变了TPR-tree的节点结构和代价目标函数；根据新的代价目标函数对插入、删除以及查询算法做出相应的改变。实验中所处理的移动对象是使用GSTD随机生成的，实验把附加属性作为索引一个维度的ATPR-tree和不作为索引的一个维度的TPR-tree进行比较。实验结果表明，对于带有属性值条件的查询，把属性值作为一个维度的ATPR-tree索引可以有效地减少节点访问次数，提高查询的效率。

关键词: 时空索引, 城市计算, 属性值, 移动对象