Research On Toponym Ontology Entity And Relation Extract

These year,the emergencies often occur. The emergency management is more and more important. Emergency management needs all kinds of data. How to offer the related data quickly and accurately is one of the problems of emergency management research. With the development of internet , data on the internet is growing exponentially. These data contains lots of information which the emergency management needs. Toponym information is the core supports of emergency information. This paper researches on extraction of toponym ontology's entities and relations to provide core base for emergency information extraction and semantic emergency information. Entity and relation extraction is the sub-task of information extraction. Existing researches usually uses the rule based methods or statistical learning based methods. According to the original text's characteristics ,this paper uses the rule based methods and statistical learning based methods separately.There is no corpus for toponym ontology extraction so this paper builds the entity and relation system of toponym ontology firstly. According to characteristics of attention in the process of extraction,this paper builds the corpus for the entity and relation extraction. This paper classifies the toponym ontology's entities and relations in the light of their occurrences in the original text. This paper summarizes the rules for some of the entities and relations to do rule based extraction and uses the Maximum Entropy for other entities and SVM for other relations. According to toponym ontology relations' characteristics,this paper builds some rules to infer implicit relations from existing relations.This paper develops the toponym ontology entity and relation extraction system and uses the data to the Chinese semantic toponym search engine. Practice has proved that the extracted toponym entities and relations which help users to get the toponym related information more conveniently, faster and more accurately have greatly enhanced the user experience.