| The complexity of software cost estimation is the fundamental reason for the lack of rationality in the budget evaluation of e-government projects.The expert evaluation method usually adopted by the government approval department will also have great differences due to the different occupations of the experts,and the realization of software cost estimation based on international standards and national standards can only rely on professionals with sufficient software technical knowledge and estimation knowledge.This is a major issue in the budget evaluation of e-government projects.Based on knowledge processing technology,this paper attempts to solve the problem based on the standard system structure of e-government information system and using the national software cost estimation standard function point method.In this paper,a knowledge ontology of software cost estimation of e-government system is established,and the content of the semi-structured software project budget table is processed by text mining technology to complete the ontology instantiation.Supports software cost machine estimates within a certain margin of error.The specific research contents are as follows:(1)This paper establishes the software cost estimation ontology of the e-government system,and expresses the function point method and the reference frame of the e-government system hierarchically.Firstly,on the basis of fully considering the measurement process of the function point method and the characteristics of the reference frame of the e-government system,the software cost estimation ontology of the e-government system is constructed,and the complex knowledge is expressed hierarchically.Secondly,in order to make full use of the application software development budget table,combined with text mining technology,a data entity extraction algorithm based on dependency syntax is designed to extract the data entities described in the project declaration file;for the functional entities in the functional description,the Bi LSTM-CRFbased The functional entity recognition model is used to extract.Finally,a method of identifying the measuring elements of function points based on knowledge reasoning is proposed,which is relatively objective and effective to identify the measuring elements.(2)On the basis of analyzing the characteristics and overall design specifications of the egovernment system,construct the software quality model of the e-government system,and use the fuzzy comprehensive evaluation method to evaluate the system quality.In this process,the LDA topic model and the FKNCN model are used respectively The judgment matrix and membership degree are calculated,and the evaluation results are used as adjustment factors to adjust the function points.Finally,by analyzing the possible model and data errors in the measurement process,training the error model,and obtaining the cost measurement results under a certain confidence interval,the software cost estimation is carried out scientifically and reasonably.(3)Based on the software cost estimation model of the e-government system proposed in this paper,using the application software development budget table in the e-government system application documents of the T City Network Information Office in 2017-2021,the software cost estimation method in this paper is specifically expounded and verified.The correctness of the method and the effectiveness in practical use are proved. |