Military Ontology Building System For Multi-source Data

Military ontology is one of the important cornerstones for constructing auxiliary military decision-making,and the main component of military ontology is the weapon ontology.There are two major sources of current weapon knowledge,one is open military knowledge websites,and the other is manual collection of combed knowledge,including databases and documents.This thesis studies the methods and tools for extracting military ontology from a variety of data sources,and designs and implements corresponding systems to efficiently construct military ontology.The main researches of this thesis includes:(1)Study the engineering scheme for constructing weapon ontology from multi-source data;(2)Analyze how to complete the ontology construction scheme through the similarity calculation during construction process;(3)Analyze methods for automatically extracting ontology concepts and relationships from military natural texts;(4)Study the method of ontology automatic construction and detection based on ontology construction principles and knowledge standard;(5)Study how to improve the efficiency and quality of construction through human-computer interaction when constructing the weapon ontology;(6)Design and implement a system that can construct a weapon ontology from multiple source data.The main contributions of this thesis are:(1)Design an ontology construction scheme for multi-source data and an ontology construction process for human-computer interaction;(2)Solved the related problems in the ontology construction scheme and realized a series of key technologies;(3)Implemented the weapon ontology construction system.In this thesis,the key technology,system function and system performance of the system are tested.The construction of the aircraft carrier ontology is taken as an example to evaluate the ontology.The results show that the system can be put into practical application.This provides a basis for further application research based on weapon ontology,and has reference significance for related research.