基于位置-文本关系的空间对象top-k查询与排序方法

针对普通的空间关键字查询通常会导致多查询结果的问题。本文提出了一种基于空间对象位置-文本相关度的top- k 查询与排序方法，用于获取与给定空间关键字查询在文本上相关且位置上相近的典型空间对象。该方法分为离线处理和在线查询处理2个阶段。在离线阶段，根据空间对象之间的位置相近性和文本相似性，度量任意一对空间对象之间的位置-文本关系紧密度。在此基础上，提出了基于概率密度的代表性空间对象选取算法，根据空间对象之间的位置-文本关系为每个代表性空间对象构建相应的空间对象序列。在线查询处理阶段，对于一个给定的空间关键字查询，利用Cosine相似度评估方法计算查询条件与代表性空间对象之间的相关度，然后使用阈值算法（threshold algorithm，TA）在预先创建的空间对象序列上快速选出top- k 个满足查询需求的典型空间对象。实验结果表明:提出的空间对象top- k 查询与排序方法能够有效地满足用户查询需求，并且具有较高的准确性、典型性和执行效率。

A location-text correlation-based top-k query and ranking approach for spatial objects

Due to the large size of spatial databases, a common spatial keyword query often leads to the problem of too many answers. To deal with this problem, this paper proposes a location-text correlation-based top- k query and ranking approach for spatial objects, which aims to find the typical spatial objects with high text relevancy and location proximity. This approach consists of offline processing and online query steps. The offline step scores the relationship between any pair of spatial objects by considering their location proximity and text similarity. Then, by using a probabilistic density-based representative spatial object selection method, a set of representatives over the spatial objects is selected to build a corresponding spatial object sequence. In the online query period, when a user issues a spatial keyword query, the location-text correlation between the query and representative objects is evaluated, and then, the top- k typical relevant objects can be expeditiously picked using the threshold algorithm (TA) algorithm over the sequences corresponding to representative spatial objects. The experiments demonstrate that the proposed top- k query and ranking approach can closely meet users’ needs, with high precision, typicality, and good performance.