Extraction and portrait of knowledge points for open learning resources

Abstract

This article explores how to use the technology of text summarization and keyword extraction to automatically extract key knowledge points from massive educational resources and use open resources to generate feature portraits of relevant knowledge points. Specifically, this article takes the field of programming competitions as an example, firstly, crawl the problem solution resources of program design related issues, use data preprocessing to clean the data, then, use unsupervised extraction models based on Bert and centrality to summarize the documents of the resources, the LDA model is used to extract keywords from the generated document summary to identify relevant knowledge points in the resource. Finally, crawl and analyze resources based on knowledge points to establish relevant feature portraits for knowledge points. Unlike manual analysis of resources, this method can automatically select candidate knowledge points, greatly reduce labor costs.