Research On Key Technologies For Geographic Information Retrieval Of Web Page Texts

In recent years,with the rise of location-based services,massive geographical information described as texts exists in the Internet.In the era of network information explosion,with the introduction of geographical constraints in search requirements,traditional Information Retrieval appears to have a functional bottleneck,which makes it difficult to discover networked information.Geographic information retrieval has became an important means for effective information acquisition.Based on existing geographic information retrieval research,this thesis takes the meteorological field as an example to conduct in-depth research on key technologies of geographic information retrieval.The specific work is as follows:(1)Based on the research of ontology-related theories,this thesis proposes a method of knowledge base construction by combining concept ontology tree and concept attribute relationship table,and builds a knowledge base with certain reasoning capabilities,which provides with a firm knowledge foundation for collection,extraction,expression,analysis and relevance calculation of subsequent networked information.The knowledge base is a lightweight knowledge base that is easy to build,apply and extend,and can meet most of the needs of geographic information retrieval.(2)Taking weather topic news as an example,this thesis develops a fixed-depth theme web crawler to collect news web pages based on web crawler technology for information collection,and proposes recognition and extraction algorithms of topic information and geographic information.At the same time,an improved triple expression model that takes care of the relationship is used to express the text information reasonably,which lays the foundation for the retrieval relevance evaluation.(3)This thesis combines the quantitative and qualitative methods to semantically understand the geographic information elements search intents and webpage texts,and focuses on the construction of semantic analysis models under topological and orientational relationships.At the same time,by introducing the concept of semantic distance,a retrieval relevance evaluation model based on the semantic distance of the concept ontology tree is proposed based on the analysis of geographic semantics,which can realize the accurate evaluation the relevance between the retrieval intention with geographical constraints and the web page text.(4)Based on the above study,a geographic information retrieval prototype system for webpage texts is developed by ASP.NET,Arc GIS API for Javascript,etc.The retrival tests of the prototype system retrieval shows that the recall rate is stable at 75%,while the precision rate is stable at 85%,which proves that the prototype system has a good search effect and verifies the feasibility and reliability of the relevant theory of geographic information retrieval in this thesis.