| In recent years,with the wide application of information technology and digital technology,online learning has become an important way of daily learning.Online learning has lowered the threshold of knowledge acquisition,improved access to educational resources,and greatly promoted educational equity.However,with the dramatic increase in the number of online courses and course resources,a number of issues have emerged that need to be addressed.First,traditional teaching methods are usually linear,following a fixed course plan,syllabus and textbook content.However,in online learning,students are more inclined to learn on their own and explore on their own.This requires a more flexible and personalized representation of course knowledge to meet the new requirements of online learning.Second,although the number of online courses and course resources has increased dramatically,their relevance to course knowledge is low and difficult to use effectively.The course resources are too scattered and numerous but not centralized,which greatly affects the learning efficiency of students.This requires the use of course knowledge to achieve effective management of course resources so that students can have better learning experiences and learning outcomes.Finally,the tasks associated with personalized learning place higher demands on the management of course resources.Course resources such as exercises and videos are the basis for tracking learning status for reasonable recommendations,which also requires a more flexible way of representing course knowledge to manage course resources.To address the above issues,the main research of this paper is as follows.(1)To address the problem of insufficient accuracy of course knowledge concept recognition,this paper proposes a course knowledge concept recognition method based on BERT-CRF fusion boundary.The method first uses BERT to obtain the embedding representation of each sentence,and then uses BERT to obtain a high-dimensional embedding vector representation that integrates the sentence context information.Then the embedding representation obtained using BERT is fed into a BiLSTM network to extract the boundary information of each knowledge concept.Finally.the output of BERT and BiLSTM is fused to obtain a joint feature representation,which is then fed into the CRF layer for decoding to obtain the corresponding label for each word,and the knowledge concept recognition of the whole sentence is completed.In addition,this paper constructs a dataset of course knowledge concept recognition based on the data structure course to measure this method against the baseline method.The experimental results show that this method is superior to the baseline method and can perform the task of course knowledge concept recognition more effectively.(2)In response to the problem of inefficient utilization of course resources and the practical need to address learning resource recommendation in adaptive learning,this paper proposes a method for linking course resources based on adversarial BERT.The method consists of two main parts,firstly,the pre-trained language model BERT is fine-tuned for solving the multi-label text classification task.Then,adversarial learning algorithms are introduced to generate adversarial samples based on multi-label classification using BERT,so that the model has generalization capability and can better handle course resources from heterogeneous sources.To verify the effectiveness of the method,this paper collects and organizes the data of exercises and blogs based on the data structure course,and constructs a dataset of course resource links.The experimental results show that this method has a significant improvement over the baseline method and can better achieve course resource linking.(3)Intelligent curriculum knowledge engineering system is designed and implemented.In order to solve the engineering problems in the process of constructing curriculum knowledge map and apply the research results,this paper designs and implements an intelligent curriculum knowledge engineering system.The system has several functions,including the annotation of course knowledge concept relations,the display of course knowledge map,the identification of course knowledge concepts,the linking of course resources and some other necessary basic functions.Through these functions,the system realizes the closed loop of course knowledge engineering. |