| With the rise of mobile devices,location-based services based on the Internet are widely used.These location-oriented services can not be separated from the information interaction between users and platforms.In fact,location is often used by human to describe information,and it is also the basis of map location service.By analyzing the semantic when people describe location,we can better collect location information and understand location scene,thus improve the accuracy of location query.However,due to the differences of spatial cognition,there is no standard rule for the semantic description of location.For example,when the selected reference object is small,people often ignore the contour range of the reference object as POI localization,and when the reference object is large,people always consider the influence of the boundary contour of the reference object on the position description.This shows that if there are complex spatial relations in the statement,such as distance,bearing,including relations,existing methods are difficult to locate the query,or query results do not conform to user semantics.These application scenarios with complex semantic descriptions put forward new requirements for localization.At present,there are many user semantic location methods at home and abroad,most of which are oriented to point-or linear-object-based semantic location information,and less consideration is given to area-object with complex attributes,features and spatial information.It is difficult to describe the position relationship between the objects because of the complex geometry of the objects,so the direct point-line location method will lead to large errors and inaccurate location results.In order to solve the problem that it is difficult to locate the area geographical entities under the complicated semantic location description,this paper,according to the characteristics of area geographical entities and the related semantic description,combines the application of geographical grid in location,this paper presents a Geo SOT grid location method for the semantic location of area geographical entities:(1)Construction of spatial semantic location database of area objects based on network data.The existing researches are mostly oriented to POI(points)or road network(lines),but there are few geographic data related to area objects.In order to realize the query and location of planar objects,firstly,the database of planar objects with spatial information should be constructed.In this paper,a large amount of geographic data is crawled from the map query API and related websites,and then,according to the semantic concepts and position information of different types of area objects,the area objects and the area objects database containing their spatial information are constructed,provides a data base for subsequent location queries.(2)Orientation segmentation of planar objects and Geo SOT grid coding computation based on nine-grid.Based on the geographic data obtained from the Internet,the object is segmented by the smallest outsourcing rectangle,so as to give the “Bearing”attribute to the segmented entity.In the segmentation process,because of the irregular shape of the area object,the orientation corresponding to the block may be blank,according to the semantic similarity,the approximate region instead.Finally,the mesh coding set corresponding to the segmented sub-region is calculated,and the transformation of the position information of the area object from coordinate to mesh is realized.(3)Geo SOT grid based method for location of area geographical entities including distance and azimuth.The method takes into account the influence of the size and contour of the area object on the location description,and uses the Chinese word segmentation technique to extract the key words such as place name,distance and direction from the location statement,the corresponding objects and trellis code sets are matched in the database,and then the semantic positions of the target objects are calculated and located by mapping the keywords to the grid algebraic relations.(4)Design multiple experiments to verify the feasibility and validity of the method.In this paper,two application scenarios are designed to locate the semantic position of the area object including distance and orientation.The experiment results show that the proposed method can realize the location of area geographic entity under complex semantics,which is a supplement to the traditional location method and has practical application significance. |
