Research And Application Of Policy Mining Based On LSTM Network

With the rapid development of new generation information technologies such as big data,cloud computing,artificial intelligence and the improvement of government informationization level,the government's governance scope,content and means have also been influential.In recent years,government big data has gradually become The focus of government managers and researchers.Through the mining of massive policy texts,on the basis of qualitative research,scientific analysis of policy texts by means of information technology and other means,to find out the personality and commonality of policy texts,and to assist policy researchers to better quantify Research and provide theoretical and technical support for policy mining research.At the same time,the policy text big data mining will have profound changes and influences on government governance.Using government big data to improve government decision-making level and governance ability is a new opportunity and challenge for government management innovation and even smart decision-making.Firstly,this paper analyses the current situation of policy mining and related literature,and combines practice and related theories and techniques to propose the mining of policy texts using LSTM network model in deep learning,and improves the LSTM algorithm model.The SimHash-LSTM algorithm is proposed for the first time.Improve the accuracy of text classification in policy mining.This study uses web crawler technology to obtain millions of policy texts from the Internet and everywhere from the Internet,and cleans the text,constructs a relatively comprehensive policy text database,and then implements policies based on the improved LSTM neural network model.The classification of texts and policy diffusion have been dig deeper,and the classification of policy texts in the smart city field and the analysis of policy diffusion examples have been provided to verify the validity of the model.At the end of the paper,a deep learning-based policy mining platform is built through object-oriented programming,which provides the entire mining process from data acquisition,data cleaning,text segmentation,text representation to text summary,text classification,similarity calculation,and topic model.Visual display.Through Django,Bootstrap framework,combined with NGINX proxy,MySQL database,etc.,the page display and interaction are realized,which makes the policy mining platform can be further applied.