Unifying Spatial Keyword Indexing in Continuous Search on Dynamic Objects
Unifying Spatial Keyword Indexing in Continuous Search on Dynamic Objects
复制标题
DOI：
10.1007/978-3-030-73103-8_29
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Y. Kato;Hanxiong Chen;K. Furuse;H. Kitagawa
中科院分区：
文献类型：
--
作者：
Y. Kato;Hanxiong Chen;K. Furuse;H. Kitagawa
人们现在使用智能手机和平板电脑等移动的设备访问Web。由于SNS和GPS的流行，移动的用户可以发送和接收关于他们的位置和他们感兴趣的内容的信息。在这项研究中，我们考虑了一个查询问题，连续监测动态空间关键字对象，其中每个查询连续监测其Top-k对象。我们引入了一个新的概念，称为排序等价。随后，在此概念的基础上，我们提出了一个新的相似度公式，统一的空间和关键字信息在同一个空间。使用我们的相似性公式，所有的对象和查询都可以在高维空间中处理，其中空间和关键字信息分别成为空间和关键字维度。在这个统一的空间中，可以预先开发一个索引来估计查询和对象之间的相似性。此外，可以使用距离公理来进行过滤技术以排除查询，避免无意义的Top-k对象更新，从而提高效率。理论分析和实验证实了我们所提出的技术的效率。
People now access the web using mobile devices such as smartphones and tablets. Because of the prevalence of SNS and GPS, mobile users can send and receive information regarding both their locations and the content they are interested in. In this study, we consider a query problem of continuously monitoring dynamic spatial keyword objects, where each query continuously monitors its Top-k objects. We introduce a new concept calledranking-equivalence. Subsequently, on the basis of this concept, we propose a new similarity formula that unifies both spatial and keyword information in the same space. Using our similarity formula, all objects and queries can be treated in a high-dimensional space, where the spatial and keyword information become the spatial and keyword dimensions, respectively. In this unified space, an index can be developed to estimate the similarity between queries and objects, in advance. In addition, it is possible to conduct the filtering technique using a distance axiom to exclude the queries avoiding meaningless updating Top-k objects, thereby improving the efficiency. Theoretical analysis and experiments confirm the efficiency of our proposed techniques.