Design And Implementation Of Industrial Economic Information Management Cloud Platform

Regional industrial economy is an important part of China's economy.In recent years,local governments have accelerated the information construction to ensure the healthy and stable development of regional industrial economy.In order to better evaluate the status of enterprises in the jurisdiction and provide corresponding services,the administrative department of industrial economy in the government needs to comprehensively collect,manage and apply the enterprise information in its area.As the management needs of regional governments for their industrial economic information are basically similar,focusing on the relevant functions and practical needs of local governments' economic management,and based on the comprehensive consideration of their business needs,this paper designs a cloud platform that can provide industrial economic information management for multiple government departments based on SaaS architecture,and realizes the main functions such as multi-tenant management and regular automatic collection of enterprise Internet information.At the same time,the enterprise information intelligent retrieval tool of cloud platform needs a large amount of training sample data,which is lack of historical data in practice.This paper designs and implements the training data generation tool according to the requirements of NLU module,which can automatically generate diversified sample data and training marks based on template.The main work of this paper is as follows.1.Design and implementation of industrial economic information management cloud platform(1)The management and application of regional industrial economic information are designed based on SaaS mode and Spring Cloud microservice architecture.According to the system business capability,service division is done and the relevant functional modules of the system are designed and completed.In addition,the Docker technology is used to deploy various services of the system.(2)The multi-tenant function of the system is completed by dynamically switching the database,making full use of the multiple leasing characteristics of SaaS,so that multi-regional governments(tenants)can share the same system to implement the economic information management in their own jurisdictions.All tenant data of the system is isolated by database separation mode to ensure data security.(3)The automatic collection of enterprise Internet information is realized based on RPA,using Python language to write script crawler,and using selenium,an automated testing tool,to simulate users to operate web pages and obtain web page data.The system monitors the change record table to update and collect external data.In case of data conflict between the data collected externally and the data collected and reported internally,manual review is provided.2.NLU training data generation(1)Generate training data based on template method and crowdsourcing mode.Through the given natural language template,SQL template and database table fields,the template data is preliminarily generated,and the entity information of the data is automatically marked.At the same time,the synonym replacement and other operations are carried out for the preliminarily generated template data after word segmentation to complete the data enhancement.Finally,crowdsourcing rewriting is used to enhance the diversity of sentence patterns.(2)In order to solve the problem that the entity information automatically labeled is wrong due to rewriting data,this paper trains the BERT-Bi LSTM-CRF model to automatically recognize the entity information in the sentence,and automatically mark after recognition.(3)The Django framework is used to realize the training data generation tool,provide template data generation,task management and data rewriting services,and celery distributed task queue is used to realize the asynchronous execution of data generation tasks.