Research And Implementation Of A Personalised Question Bank System For It Examination

With the development of internet technology,the informatization of education has become an unstoppable trend.Rapidly developing online education platforms have provided learners with abundant learning resources,and learners have also left a large amount of learning data on these platforms.How to mine this data to achieve personalized learning services and meet the needs of learners for personalized education is an important research topic in the fields of education and computer science.This research focuses on building a question bank system for the Shanghai Higher Education Information Technology Proficiency Examination,in order to provide personalized learning services for the exam candidates(i.e.,learners).Examinees’ needs are to learn the knowledge points required in the examination outline,and to identify and fill gaps in order to pass the exam.At the same time,the needs of learners at different levels are also different,so it is necessary to model learners’ knowledge states and design personalized learning recommendation models around the above needs in order to provide personalized recommendation services.The core of personalized learning recommendation technology is the knowledge level diagnosis method,and knowledge tracing is currently one of the mainstream methods.However,some knowledge points in the question bank of the Information Technology Level Examination currently have the problem of being too coarsely divided in granularity,which cannot meet the requirements of the knowledge tracing model very well.Therefore,it is necessary to cluster the questions under knowledge points that have been divided too coarsely in order to achieve the purpose of further subdividing knowledge points.The main contributions of this research are as follows:(1)This article proposes a clustering scheme for exam questions based on the STCKH model,which aims to address the problem of coarse-grained knowledge point division in the information technology proficiency exam question bank.The model clusters questions under coarse-grained knowledge points to achieve further division of knowledge points.The model is based on the KACLSE text embedding model and the HDBSCAN algorithm.The KACLSE model proposed in this article improves text embedding quality through contrastive learning,which helps the STCKH model to perform text clustering more accurately.Experimental results show that our method achieves higher accuracy in clustering texts under each first-level category compared to the baseline model.The clustering results of exam questions will provide relatively accurate knowledge components for knowledge tracking models involved in personalized learning recommendation tasks.(2)In response to the personalized learning needs of learners in the information technology proficiency exam question bank system,this research proposes a personalized question recommendation model PTRMP.The PTRMP model is based on the knowledge tracking model PCBKTF proposed in this research to predict learners’ mastery of knowledge points,and recommend appropriate difficulty questions to enhance learners’ engagement.At the same time,the model can balance learners’ needs in filling gaps and exploring new knowledge.The PCBKTF model improves the BKT model’s inability to capture the recency effect and insensitivity to order,as well as the problem of not considering differences in learners’ individual abilities,by simultaneously introducing a forgetting factor and clustering learners into groups,thus improving AUC.Relevant online and offline experiments show that the PTRMP model has higher accuracy and coverage than the baseline model.(3)In order to put the personalized question recommendation model PTRMP into practical use,this research designs and implements an online question practice system.The system adopts a B/S architecture,with good stability and a user-friendly and concise human-computer interaction interface.The system’s user end has functions such as question practice,exam simulation,and review,and its core function is to use the PTRMP model to analyze learners’ learning logs and provide personalized question recommendation services to help learners efficiently master knowledge points and make up for deficiencies.In summary,this research proposes a personalized test question recommendation model,PTRMP,based on the needs of learners participating in the Information Technology Level Examination.To address the problem of some knowledge points in the test question bank being too coarsely divided to meet the requirements of the knowledge tracking model,this research proposes a test question clustering scheme based on short text clustering.Experimental results demonstrate that the personalized test question recommendation model PTRMP and the short text clustering model STCKH can better meet the requirements of the background of this research compared to other methods.