Design And Implementation Of Document Management System For Government Work

In recent years,with the development of Internet technology,government staff are facing the need to analyze,manage and write a large number of working documents in the form of electronic document in a short time.How to save time and effort from all kinds of government work documents quickly and accurately obtain the synoptic information of documents,manage and edit documents,is the main problem to be solved.In order to solve the above problems,this paper combines the characteristics of government work documents,designs and implements a document management system for government work by using natural language processing technology and system front-end and back-end development technology.This paper mainly does the following work.Firstly,according to the requirement that staff need to quickly acquire and browse the document profile,the document summary and keyword are used to form a synoptic description of the document,which is convenient for the staff to quickly understand the core content of the document.First,a method based on GAN is used to realize the function of text summary,which solves the problem that the abstractive summary needs a lot of data annotation and the titles of many government documents cannot express the general information;Then,in order to optimize the summary model,the text classification function is implemented by using BERT model to classify the input data,and classifying training is used to improve the model effect;On the other hand,the function of keywords extraction is realized by Text Rank algorithm.In order to optimize the algorithm,Word2 vec technology is used to calculate the correlation between words,and the algorithm effect is improved.Secondly,the semantic search function of documents is realized according to the requirement that the staff need to get the content of semantic related documents quickly.In order to simplify the workflow of staff,searching semantic similar document in the form of document object and document content is realized,using text matching method based on Sentence-BERT model.While retaining the excellent effect of the BERT model,the problem of multiple text matching taking a long time is solved.Thirdly,according to the staff need to correct the Chinese text,the text correction function is realized.Considering that the errors in the actual writing of working documents are mainly wrong pronunciation and shape or special words,and considering the algorithm execution efficiency,the algorithm is based on the statistical language model N-gram,and the user-defined confusion set is added as the auxiliary.On the premise of meeting the system execution efficiency,the scalability is improved.In the form of the system,according to the habit of government staff using Word software for office work,human-computer interaction is provided by the combination of Word software plug-in and website,the system uses Spring Boot,My SQL and other technologies for the back-end system,uses Flask micro-service publishing and calling to integrate underlying algorithm.The system includes six functional modules: user management,data processing,document synoptic information acquisition,document semantic search,document editing,and data management.On the basis of providing document management function,it can help users obtain document category,summary,keywords and other synoptic information,carry out document semantic search,and automatically correct the text errors.At present,the document management system for government work has been tested and runs well.It can make the government staff manage and edit documents very convenient,reduces the workload of staff,and it has strong practicability and usability.