Construction And Application Of Multi-source Geographic Information Map Supporting Compound Condition Query

Map applications can provide users with geographic information retrieval services,and are essential tools for people’s daily travel.As the scenarios that need to use map applications for query become more and more abundant,users have more diversified and complex requirements for geographic information retrieval services.It is hoped that map applications can support querying by combining multiple geographic attribute conditions or spatial relationship conditions.but the current map applications on the market can only use a single condition to query geographic information.In order to solve this problem,this paper collects data from four map manufacturers,aligns the data pointing to the same geographic entity in reality,and then constructs a geographic information map by establishing the spatial relationship between geographic entities.The graph has rich geographic entity attributes and spatial relationship information,which can well support the complex conditional query function of the geographic information system.The main research and implementation of this paper are as follows:(1)For the multi-source geographic entity alignment problem,based on the FEBRL entity alignment algorithm,this paper designs a SI-FEBRL(Spatial Index FEBRL)entity alignment algorithm for geographic entities.The algorithm makes full use of the spatial characteristics of geographic data,and partitions all geographic entities by creating a spatial index,which greatly reduces the scale of candidate entity pairs.At the same time,the similarity measurement method was improved according to the spatial characteristics of geographic entities,and the ensemble learning classifier XGBoost model was trained to classify candidate entity pairs,and the aligned geographic entities were screened out.Finally,a comparative experiment was conducted on Gaode and Baidu map datasets.Compared with the FEBRL algorithm and the Dedupe algorithm,the SI-FEBRL algorithm has a 12.5%improvement in the F-measure index,which proves that the algorithm can better complete the geographic entity alignment work.(2)For the construction of geographic information map,due to the large scale of the geographic entity set that integrates the data of the four map manufacturers,all spatial relationships between geographic entities are established by the method of serial spatial relationship calculation,and the computational cost is unbearable.Therefore,this paper designs a method for constructing geographic information maps through parallel spatial computing.The method is based on the Hadoop cluster big data processing framework,uses Geohash technology to encode and partition all geographic data,which greatly reduces the spatial computing operations required to build the map,and then uses the distributed file system HDFS to store all data,And through the MapReduce parallel programming framework,the parallel calculation of the spatial relationship between all geographic entities is realized,and the geographic information map is successfully constructed.Finally,the comparative experiment proves that the parallel construction method is more than 4 times more efficient than the serial construction method.(3)Taking the constructed multi-source geographic information map as the data base,this paper designs and implements a geographic information system that supports compound conditional query.The system can be divided into data management module,compound condition query module and other map-related function modules.For multisource geographic data,through database design,the graph database Neo4j is used to store the geographic information map,as a supplement to the map data,and the spatial database PostGIS is used to additionally store all the geographic data;For the compound conditional query function,a set of compound conditional query rules are defined,For the compound conditional query function,a set of compound conditional query rules are defined,Thereby,the optimal query result can be returned to the user in the shortest time;For other map-related functions,it mainly includes functions such as geometric space collection,people flow map query,and user personal center,which can provide users with a better experience.Finally,the function and performance of the system are tested respectively,and it is verified that the system can perform compound conditional query correctly,and the main functions are feasible and reliable.