Research On Gazetteer Information Retrieval Service Based On Spatial Semantic Computation

Gazetteer information retrieval is a specific research direction in the field of geographic information retrieval.Gazetteer;refers to the names of geographical things,including formal names and aliases used by people in daily life.The related research results can meet the demands of the individuals,enterprises and governments,which are widely used in decision analysis of the daily life,traffic scheduling,resource planning and other fields.But at present,with the development of Internet technology and the continuous growth of information,information service has gradually turned to knowledge service.The traditional gazetteer information retrieval based on keyword query has been unable to meet the growing practical needs of people.It is urgent to carry out research on place-name information retrieval service based on spatial semantic correlation.For the space semantic computation of spatial objects,we construct the gazetteer retrieval model involving spatial relationship semantics,enriching the original query methods.On this basis,we further explore the latent semantic association between gazetteer objects and propose the retrieval algorithm of Region-of-Interest based on space semantic,expanding the connotation of gazetteer information retrieval service.Our main contributions include:1.the gazetteer retrieval model involving spatial relationship semantics.In this thesis,considering the current gazetteer retrieval method mainly based on keyword query,we change the retrieval mode structuring and optimize the inquiry procedure.By introducing the semantic similarity of spatial relation,we enrich the original gazetteer retrieval model.The experiment result shows that our method in this thesis can improve the query performance and make the query method more flexible.2.the correlation analysis of potential spatial semantic between geographical name objects.In this thesis,the existing representation learning method is adopted to model the distributed representation of spatial semantics between geographical name objects and to explore the potential semantic associations between objects.The experiment result shows that our proposal can effectively capture the association of spatial characteristics between objects,making their representations contain valuable semantic information.3.the Region-of-Interest retrieval algorithm based on the distributed representation of spatial semantics.We reorganize geographical name objects based on distributed representation,and realize the generation and optimization of candidate Region-of-Interest.On this basis,we design the novel query modes for Region-of-Interest retrieval.It is proved that our method can effectively discover the target Region-of-Interest on the real data set,which provides a new query strategy for the place name retrieval service.