Design And Construction Of Intelligent Question Bank

With the country’s vigorous advocacy of education informatization,the research of online question bank and other teaching assistant systems has become a hot topic at the moment.The traditional question bank only digitizes the knowledge in the books,without effectively using teachers’ knowledge reserves and students’ question data,ignoring students’ personalized learning needs.In view of the problems of traditional question banks,there have been many researches on smart question banks in China,but most of them only focus on students’ needs and are dedicated to achieving accurate question recommendations and personalized group papers,without taking into account the needs of teachers and the expansion of interactivity among teachers,students and question bank systems.This paper explores the needs of students,teachers,and the system as well as the interactivity among the three to create a trinity of teacher + student + system for an intelligent teaching ecology.The paper implements the following work:First,starting from the needs of students,propose a set of personalized recommendations based on the similarity of topics.Analyze the characteristics of users and the question bank,classify the data into 4 dimensions of question similarity,question popularity,user interest,and user ability matching,analyze the characteristics of different dimensions of data one by one,customize personalized collection schemes,and realize timely response to recommend accurate personalized exercises function.Second,starting from the needs of teachers,realize the function of importing highly intel igent questions.Crawl and collect basic computer knowledge corpus by itself,construct a topic self-categorization model based on the BERT pre-training model,and realize the self-categorization function of imported topics.Design and implement a complete set of question import function system from question file inspection,duplicate check,self-categorization to manual verification.The topic selfcategorization model and the teacher support each other.The model assists teachers to upload topics in batches,and the new topic and the model classification results verified by the teacher are used for the model to further learn domain knowledge.The system can automatically process newly added topics into model corpus for incremental model update and realize autonomous evolution of the model.Third,build a feature-rich and interactive question bank system.The question bank client has the functions of exercise practice,review,exam simulation,personalized recommendation and built-in text review forum;the question bank management side supports the update of the self-classification model of questions and the WMD model used for question similarity calculation,as well as the manual and automatic dual-mode update of pre-stored data required in the personalized recommendation function.