Research On Spatial-Textual Index

Spatial information is one of the most important information resources. Spatial information system on internet is the main provider of dynamic spatial information. Studies on spatial information search engine in geographic information services and share have practical significance. Geographic information retrieval which is used in spatial information search engine, as a traditional branch, including all the traditional core areas of information retrieval, but more concerned about the spatial indexing and retrieval. Geographic information retrieval display geospatial or object in geographic space, and deal with one or more spatial information related to multiple place. Geographic information retrieval technology treat geo-referenced information in text as a rich data source. Through analysis and usage of text, providing users with more geographic information services. Indexing methods are an important aspect of geographic information retrieval. In this paper, the index methods of domestic and international geographic information retrieval is studied. The indexing methods which retrieve documents from documents collection by geographic revelant is studied, considering the text information and geographic information in the documents, using placename identifying methods based on conditional random field and spatial relation identifying methods, the main contents are as followes:(1)Text spatial semantic computation model, according to the frequency of placenames, spatial relation and spatial distribution, the factors that affect the spatial semantic computation model are analyzed. In this paper, a model is proposed which analyse the spatial semantic correlation; and context correlation to get the theme correlation.(2)Spatial-textual hybrid index based on R-tree. Based on the analysis of text indexing and spatial indexing methods, explore the merging methods of two indexing methods. Three types of indexing mechanisms is built and implemented in our experiments:spatial priority indexing method, text priority indexing method, and spatial-textual double indexing method. The storage and efficiency are compared in our experiments.(3)Development of a prototype system and case study..A prototype system of spatial-textual indexing method has been designed and developed.The system includes placename identification module, spatial relation extraction module, space calculation module and spatial-textual indexing module.