Research On Curriculum Knowledge Extraction And Its Application

The rapid development of artificial intelligence has had a tremendous impact on many industries.Natural language processing,a hot research direction in artificial intelligence,has been widely used in machine translation,intelligent question answering,and speech recognition.Intelligent Education is a field that combines artificial intelligence and natural language processing technology with traditional educational means to explore new modes of education.Currently,most of the research on Intelligent Education focuses on digitizing traditional education resources and exploring them as a database.This practice has played a particular role in promoting the development of Intelligent Education,but there are still some defects.The most fundamental problem is that the organizational structure of digital educational resources can not meet the needs of the upper application of Intelligent Education.Because of this problem,this paper proposes to combine Knowledge Graph with digital education resources,and at the same time,build the overall framework of Intelligent Education based on atlas data,and conduct in-depth research,design,and implementation of each level in the framework.At last,we design some experiments to verify the methods in this paper.The experimental results prove the effectiveness of the overall architecture and methods.The research work of this paper mainly includes the following three aspects:1)In order to solve the problem of structuring curriculum knowledge according to the information extraction theory,combined with the suitable methods of Chinese information extraction and professional field information extraction,this paper proposes the overall framework of curriculum knowledge extraction and application.It defines in detail all research and experimental methods that need to be studied from the data bottom layer to the actual application layer.Firstly,the framework constructs a Chinese course dataset by combining manual screening with remote monitoring;Then,based on using the Bert pre-training model to encode the text,combined with the dependent syntax tree of the sentence,the graph convolution network is used to learn the word and semantic information of the sentence.Then the relational attention is superimposed to learn the relational features in the sentence to extract and encode the sentence features.The LSTM is used to decode the encoded information.Finally,the extraction of entity relations is completed to form structured triplet information.The unstructured text information is initially converted into structured information.Experiments show that this method has achieved good results in Chinese information extraction and curriculum knowledge extraction tasks.2)Because of the flexible application of curriculum knowledge triples in Intelligent Education,this paper first completes the attributes of curriculum knowledge triples,crawls the attributes of entities in the Knowledge Graph using web crawler technology,and dramatically expands the essential attribute information missing from the triples proposed from the curriculum text,making the entity information in the atlas more sufficient.At the same time,the graph database neo4 j is used to store the Knowledge Graph,which provides a database for the practical application of course knowledge.In order to further optimize the expression effect of the Knowledge Graph,this paper also makes relevant improvements to the classical algorithm Page Rank so that it has the ability of weight attenuation to solve the defect that the weight calculated by this algorithm is related to the number of links.At the same time,the weight calculation method of the relationship pattern is defined.In order to visually analyze the above research contents,the graph layout method based on force oriented algorithm is also explored.3)Complete the relevant exploration and attempt of the landing of Knowledge Graph in the field of Intelligent Education.In order to assist the complete education of Knowledge Graph,this paper studies cognitive diagnosis methods and knowledge recommendation algorithms and proposes a test-based cognitive diagnosis method for learners TALCD and a recommendation algorithm CERRA based on entity relationship intimacy.TALCD is based on a series of predefined sets and scores and puts forward the threshold of mastery and retention coefficient of mastery of knowledge points.It evaluates the learners’ mastery of knowledge points through the answers to questions;CERRA,based on the results of TALCD,flexibly recommends knowledge points that need to be learned or reviewed to learners through their mastery of specific knowledge points.This paper designs and implements a personalized knowledge recommendation system based on curriculum Knowledge Graph,TALCD,and CERRA and explores a new method of applying Knowledge Graph in Intelligent Education.The experimental results show that the method proposed in this paper has some advantages in the field of Chinese information extraction,especially in the field of curriculum knowledge extraction;It is entirely feasible to apply curriculum knowledge to Intelligent Education in the form of a Knowledge Graph.The research in this paper has a positive role in promoting the application of curriculum knowledge in Intelligent Education.