Data-driven Computer Curriculum System Analysis

With the rapid development of modern computer technology,in today's social production and life,all walks of life need computer technology for technical support,and use computer technology to reform the previous production technology,so as to improve product quality and reduce production costs.Therefore,the sustainable development of computer technology and the cultivation of computer technology talents suitable for the new century has become one of the important factors for the sustainable development of society.Throughout the source distribution of computer talents in the society,the vast maj ority of high-quality computer talents are produced through the cultivation of undergraduates,so the cultivation of computer talents in undergraduates is very important.The cultivation of undergraduate talents,need to develop a good syllabus,this artical puts forward a set of effective way,using computer technology to process the syllabus document,to assist the syllabus maker to develop the syllabus document.In order to have a better reference,this artical also selects ACM and IEEE Computer Society issued "computer science curriculum system specification 2013"as the measurement standard of syllabus document.The main work of this artical is as follows(1)In this artical,we collected the data of undergraduate course syllabus of computer science and technology in some domestic universities,and provided it to other researchers for research;(2)This artical proposes a set of process flow of course description information and a userdefined similarity calculation method between courses.After the course description information is segmented,TF-IDF,textrank,LSI and LDA are used to quantify the course description information,and the distance measurement between vectors is used to calculate the similarity between courses.After the segmentation result is quantized by corpus,the user-defined maximum similarity method is used Secondly,the similarity of course description information is calculated by formula;(3)This artical proposes a new clustering algorithm.After the word segmentation results are vectorized by corpus,in order to speed up the calculation,K-means and k-medoids algorithms are used to cluster the word vectors.Based on the idea of k-medoids,this artical proposes a k-center selection method based on simulated annealing algorithm,which uses the clustering results and user-defined maximum similarity method to calculate the distance between the course description information Similarity.The effect of the proposed clustering method is close to that of K-means method,but the result has more obvious semantic analysis than k-means method,and is better than k-medoids algorithm in time complexity and result evaluation index.(4)This artical proposes a method to calculate the similarity between user-defined syllabus files.After getting the similarity results,the user-defined maximum similarity method is used to calculate the similarity between syllabus files,so as to help the syllabus file makers to make syllabus files.