A Semantic Expansion Model For Geographical VGI Information Retrieval

In recent years,with the developing of crowdsourcing data,volunteered geographic information is growing explosively.OpenStreetMap is one of the most representative projects among VGIs,and its data is widely used in various fields.The spatial information retrieval is becoming more and more important.While the proportion of non-professionals in this search activity has increased significantly,users generally use simple and easy way for information retrieval,but the actual search results often deviated from the user's expection.In order to make the search results more accurate,people have further constructed combined keywords to achieve better search performance.But most users do not have retrieved training experience,it is difficult to make good combination of retrieval methods to construct complex search strategies.How to provide users with a simple information retrieval method,as well as a good search performance is widely concerned in academic circles at home and abroad.There are many problems in the acquisition,querying and retrieval of spatial information due to the characteristics of geo-data,include the big data problem,the heterogeneity of geo-data service,the ambiguity of user demand expression and so on.Spatial information retrieval has significant spatio-temporal characteristics,it is a special application of information retrieval.The traditional information retrieval technology based on keyword string matching can not fully meet these requirements,which needs a better solution such as semantic-based information retrieval and matching.Semantic method,as a new type of knowledge organization and expression,has a good conceptual structure for logical reasoning.It has been widely used in information retrieval field,especially in knowledge-based semantic retrieval.Semantic based description and reasoning method provides a feasible solution for the key technical problems such as analysis and understanding of user requirements in spatial data service,semantic description and matching of spatial information resources,formal description and reasoning analysis of domain knowledge and so on.It can effectively improve the efficiency of retrieval results and intelligence of services.This paper designs a user query intention recognition method for the natural language processing of spatial information retrieval statements,which is based on the query of the spatial information retrieval users.In addition,using the tags in OSM data,combined with spatial information semantic features,the attribute information items and values can expanded in the tag values.An OSM ontology model with semantic property items is constructed and a query expansion method is proposed based on the similarity of properties of the ontology model.Moreover,a relevant experiment is conducted with the OSM data of China.The experimental results demonstrate the effectiveness and efficiency of the proposed method for geographic information retrieval.This study provides a method and example that can be used as a reference for subsequent research on spatial information retrieval.The study mainly includes the following aspects:(1)By constructing the spatial information and gazetteer semantic dictionaries,it can make better performance in query sentences splitting and query intention understanding.Because of the semantic dictionaries support,it is more accurate to identify the geo-names and avoid the occurrence of query offset.(2)Effective use of VGI tag scalability,can greatly enrich the spatial data semantic features,and provide a good attribute expanding platform.Spatial data semantic attribute extension make spatial data deep analysis and mining possible.(3)By using the semantic-based spatial information similarity query extension model and method,it can make better performance in spatial information retrieval,and greatly improving the efficiency of retrieval results.(4)The experiments of this paper proved the feasibility and efficiency of the semantic expansion model for geographical VGI information retrieval.This study provides a method and example that can be used as a reference for subsequent research on spatial information retrieval.