Research On Knowledge Graph Construction Method For Ship Command Decision

With the widespread use of the Internet,the amount of information that comes with it has gradually increased.The knowledge graph has attracted scholars’ attention because of its utility in semantic processing and organizing data.In view of the large amount of heterogeneous and multivariate data stored in ship command,it is urgent to integrate the organization.The traditional method cannot well organize the knowledge.To this end,the thesis proposes to construct a knowledge map for ship command decision-making in order to Integrate knowledge of ship command on the battlefield from existing network data.Extracting situation elements from a large amount of multivariate heterogeneous data and performing entity extraction are indispensable operations in constructing a knowledge map.The thesis mainly uses the situation element extraction method in conjunction with artificial neural network models to perform entity extraction of situation elements.According to Existing methods of constructing maps acquire knowledge and perform knowledge reasoning on the knowledge,thereby expanding more valuable information about ships.The research focus of this thesis mainly focuses on the two parts of entity extraction and relationship extraction from data.The research work is as follows:First,in order to extract entities more accurately and effectively,a method of entity extraction based on weighted PNN and feature matching is proposed.Aiming at the problem of extracting the relevant entities of the ship command in the network data,the current popular entity extraction method is used to analyze the ship command data for the entity extraction.The extraction process does not combine the unique situation elements of the ship to perform the extraction.One problem,using the element matching method,a semi-supervised entity extraction method based on element matching is proposed,which can improve the accuracy and accuracy of entity extraction.In addition,for element matching,situation element acquisition is also required.Analysis of existing situation element acquisition methods reveals that there is still a problem of poor classification of situation elements.In order to solve this problem,a sensitivity analysis method is used to integrate into the PNN structure.A novel situation element acquisition based on weighted PNN is proposed.Secondly,in order to better construct the knowledge map of ship command to assist decision-making,a relationship extraction method based on PGM and PSO clustering was proposed.Aiming at the problem of the relationship extraction method for ship command,the accuracy and accuracy of relationship extraction are not high.The possible problems of the extraction method are analyzed.It is found that the popular clustering algorithm based on PSO has a slow convergence rate and is easy to fall into the local optimal Solution.In order to solve this problem,a fitness function optimization method was adopted,and a PSO clustering algorithm based on fitness function optimization was proposed.It can converge faster on clustering algorithm based on PSO and is less prone to localization.Optimal solution.In addition,for the problem of low recognition accuracy of entity pairs,analyzing the characteristics of the entities,it was found that some low-similarity entity pairs were eliminated during recognition,which caused some existing entity pairs to be filtered out.In order to solve this problem,The problem is that PGM is used to match entities with entities.This matching does not filter low similarities,but matches them all;then calculates similarities to eliminate ambiguity.Finally,the method proposed in the thesis is tested with existing methods,which has higher extraction accuracy and accuracy.These methods are also applied to build a knowledge map for ship command and decision.