Design And Development Of Reporting System Based On Question Answering System

With the continuous development of information technology,data already exists in various industries and has become an important production resource.The report system can display data in multiple dimensions and provide data analysis and advice for production activities in various industries.However,the reporting system has its problems.There are many types of charts in the report,a large number of charts are stacked,and the arrangement between the charts lacks correlation.Due to the irrationality of data display and the understanding threshold set by users for complex and diverse chart types,it is difficult for users to quickly obtain demand information from reports.Users want to quickly obtain the required information from the report,and this requires the report system to quickly process the data,accurately locate the data,and display the data reasonably and efficiently.In order to solve the problems in today's report system,this article designs a report system based on question and answer.The system combines a reporting system with a question and answer system.The question and answer system communicates with users in the form of question and answer,understands user semantics,and translates user semantics into database instructions.The reporting system visually displays the data returned after the database executes the instructions.The main innovations of this article are as follows:First of all,in order to realize the conversion of natural sentences into SQL statements(NL2SQL),this paper designs a limited domain question answering algorithm module,which outputs database instructions that the computer can understand and execute.In order to avoid the equivalent serialization problem in the traditional sequence-to-sequence(Seq2Seq)model,the question and answer module uses the concept of slot filling and task decoupling to process NL2SQL tasks,convert text generation tasks into classification tasks,and improve the system Robustness.In addition,in order to avoid the result error caused by the polysemy problem of words,the module also introduces Bert-wwm preprocessing language model to strengthen the representation of context information in natural language.Secondly,this article implements an intelligent report display module.After the question and answer module generates a database query SQL statement,the system obtains user demand data by executing the database query statement.The intelligent report display module selects the appropriate chart type for the data according to the attributes of the data and displays the chart to the user in a proper arrangement.Finally,based on the above modules,this paper implements a report system based on question and answer.The system uses the Vue+Spring Boot+MySQL technology stack to achieve,the front end uses the Vue framework to implement the user interaction interface,the background uses the Spring Boot framework to process user requests,and the database uses the MySQL database to store knowledge database data.