Design And Implementation Of Knowledge Graph For Financial Courses

In recent years,with the gradual deepening of concepts such as education informatization and personalized education,autonomous learning on the Internet has become one of the important ways for students to acquire knowledge.The online education platform will provide students with different forms of knowledge learning methods,such as knowledge dissemination in the form of videos and documents.However,because the teaching resource data in the platform are independent of each other,they are scattered in the web page structure,and the learning resources are not connected through logical relationships,and it is difficult to form a systematic knowledge system structure.This problem causes the rich semantic information hidden behind the course information to be ignored,and the usable value of the course data is greatly reduced.In response to the above questions,this thesis uses python language,based on open field web data sources,designs and develops a financial curriculum knowledge graph,and realizes the systematic construction of financial curriculum information.The specific work content of the thesis is as follows:(1)Analyzed the targeted problems and objectives of the knowledge map of financial courses,and listed the key technical points to be solved.According to the guidance of the project demand analysis,the construction process of the knowledge map of financial courses was designed,including data source acquisition.,Data preprocessing,financial course ontology model design and establishment,financial course corpus keyword extraction,information extraction(entity and relationship extraction),and Neo4j-based data storage and display.(2)Data source acquisition and preprocessing operations.Design the paging data crawling process and method,use web crawler technology to crawl the course attribute information(course name,teaching institution,teaching teacher,course overview,prerequisite courses)in the Mooc website of China University,and perform the initial corpus The necessary preprocessing operations,including data cleaning operations for null items,over-long items,and sticky items,and subsequent word segmentation processing,separate a better sample corpus on the basis of the original data.(3)The design and establishment of the ontology model of finance courses.The financial course ontology model is designed and described from three dimensions: semantic relationship,semantic type and ontology object.From these three dimensions,the semantic relationships in the knowledge map of financial courses can be divided into four categories,namely: the upper-lower relationship of the upper-level classification of the course,the relationship of belonging,the relationship of professors and the relationship of prerequisites.Semantic types can also be divided into four categories: course entity,institution entity,label entity and person entity.And on the basis of the model,the ontology objects of financial courses are instantiated and constructed.(4)The realization of keyword extraction and knowledge storage module of financial courses corpus.Using keyword extraction algorithms based on TF-IDF and information entropy,respectively,the text type data contained in the course overview attributes are used to extract keywords related to finance majors,and the extracted keyword sets are used as the subsequent field named entity recognition model External dictionaries to assist in automated sequence labeling tasks.And realized the knowledge storage module that converts the structured data in the relational database into triple data and stores it in the graph database for relational linking and persistent preservation,and the storage results are displayed.(5)Financial course entity recognition method based on Bi LSTM-CRF algorithm.First,a method for automatic labeling of the original unlabeled corpus was designed and related work was implemented.Then,the labeled corpus of financial courses was used as the training data set of the named entity recognition model based on the Bi LSTM-CRF algorithm,and the financial courses were trained.Entity recognition model,then a comparative experimental analysis of the construction effect,and applied to the extraction of prerequisite relations in the course.