基于对象集合的空间关键词查询

在进行空间关键词查询时，有时需要查找一组既紧凑且离查询点最近、又覆盖查询关键词且对象个数很少的对象，而现有的查询方法通常只能返回包含所有查询关键词的单个空间对象。为此，提出了解决此类查询问题的近似查询算法和精确查询算法。首先给出了这类查询问题的形式化定义，以及描述对象集合质量的代价函数，并对代价函数进行了归一化处理；然后在近似查询算法中采用基于IR-tree的最佳优先搜索策略进行剪枝，有效缩减了查询候选空间；在精确查询算法中采用基于IR-tree的广度优先搜索策略查找包含查询关键词的对象，以达到降低查询处理代价的目的。实验结果表明，近似算法的查询效率明显优于精确算法，且能获得非常精确的查询结果。

Spatial keyword query based on collective object

In spatial keyword query, a group of objects involving minimum number that cover the query keywords and are compact and nearest to the query location will be queried, but generally, the current query method only can return the single spatial object containing all query keywords. To address this query, an approximate algorithm and an exact algorithm were introduced. The query problem was formally defined. A new cost function to describe the quality of the collective objects was defined and normalized. Then the approximate algorithm utilized a best-first search based on the IR-tree to prune the search space. The exact algorithm utilized a breadth-first search based on the IR-tree to retrieve the objects that contain some query keywords to reduce query processing cost. The experiments show that query efficiency of the approximate algorithm is much better than that of the exact algorithm, and approximate algorithm is capable of achieving very accurate results.