Research On Multi-source POI Information Fusion And Credibility Evaluation Technology

With the rapid development of new technologies such as the Internet and cloud computing and the rapid update of mobile smart devices,network geographic information services have gradually integrated into people's lives and learning.Geographic information has become an important basis and basic information for the daily life of the public,business operations,government actions,etc.Various geographic information service providers have accumulated a large amount of data resources,but the data are isolated from each other and it is difficult to integrate and utilize them.With the urgent needs of data applications,the use of the Internet's massive geographic information resources to give full play to the potential value of open source geographic information and social geographic information has received more and more attention.POI is known as the freshest "blood" of network geographic information,and it is also the basis of location services and navigation.People store,manage and analyze data of varying quality from different information sources in the military,business,environment,resources and other fields,and have a deeper understanding of the importance of data fusion.How to use multi-source POI data to effectively integrate it has gradually become one of the hotspots in geographic information technology research.This paper proposes a multi-source POI information consistent classification extraction and a multi-source POI information matching and fusion method,and evaluates the credibility of the fusion results,which provides technical support for the development of geographic information data applications.Mainly include the following aspects:(1)A consistent classification and extraction method of multi-source POI information based on name and type.Perform synonym expansion on the type of certain extraction,distinguish the expanded synonyms,and extract keywords from them;use keywords to perform fuzzy query on POI information,and query POI information containing keywords.Calculate the similarity of the query results.The POI information is sorted in descending order using similarity,and the incorrectly extracted POI information is manually eliminated.(2)Multi-source POI information matching and fusion method combining non-spatial information and location attribute information.Filter records within a certain range in the benchmark data by latitude and longitude.The administrative divisions in the POI information address are eliminated level by level according to the order of provinces,cities and counties.Calculate the similarity of the name and address by combining the edit distance and cosine similarity,calculate the spatial distance using the semi-positive vector formula,and then use the Gaussian function to perform inverse distance weighting to obtain the spatial distance similarity,and comprehensively calculate the comprehensive similarity of POI information,After setting the threshold,the point pair with the same name is higher than the threshold and has the highest similarity.Finally,merge the information of the same name.(3)Internet POI fusion entity credibility calculation method.Through the statistics and analysis of POI fusion entity name features,a semantic feature library is established.Use the semantic feature library to query the frequency of feature words and the frequency of part-ofspeech combination patterns of entity names,and calculate the credibility of the name.Comprehensively consider the number of information sources,attribute consistency,attribute integrity,and number of address elements of the fusion information,and calculate the data support.Use the search engine to search the fusion entity on the network,analyze the number and results of the search,and realize the network credibility calculation based on the search engine.Comprehensive calculation to obtain the comprehensive credibility of the POI fusion entity.Experimental results show that this paper uses similarity combining edit distance and cosine similarity to be feasible for multi-source POI information.The credibility model proposed in this paper can effectively calculate the credibility of the POI fusion entity,has achieved good results,and can provide effective help in data verification.