Research And Application Of Petition Data Based On Bert-BiGRU

With the deepening of the "three turns" work of the discipline inspection and supervision system,many problems,such as the high intensity of petition work,many ideas,strong timeliness,the shortage of petition cadres,the ability to be improved and so on,have become the main bottleneck restricting the development of petition work and need to be solved.In order to meet the needs of the current work,it has become an urgent need to deal with the petition affairs by means of information technology.The intelligent management system of petition data is based on the petition data of the provincial Commission for Discipline Inspection.It uses data mining,machine learning,text classification and other information processing technologies to establish an empirical model,generate data visualization petition statistical reports,and provide services for discipline inspection workers and decision makers.There are two aspects in this paper:(1)Establish the empirical model of petition data analysis.Based on the characteristics of petition data,this paper proposes a Bert-BiGRU network model,which uses TF-IDF algorithm to complete the text classification and keyword extraction of petition data.Firstly,the preprocessed data is input into the Bert pre training model to get the formatted text vector;Then,the text features are extracted by BiGRU,and the Attention self attention mechanism is used to sum each time sequence vector output by BiGRU to optimize the feature vector;Finally,TF-IDF algorithm is used to complete text classification and keyword extraction.(2)Scientific statistics of petition data,design and implementation of the petition data management system.The results of text classification and keyword extraction are used to mine the valuable information in the petition data through Softmax layer to generate the petition statistical report.Springboot and React technology are used to build B/S network platform.The report of petition statistics is displayed dynamically in real time through the chart plug-in of Echarts,which can be used by petition workers and decision makers.By inputting the petition data into the Bert-BiGRU model,the classification results show that the model is superior to the traditional recurrent neural network model in both efficiency and accuracy.The petition data management system constructed in this paper improves the core competitiveness and business innovation level of the petition work of the Discipline Inspection Commission.