Font Size: a A A

Research On Auto-Evaluation Method Of Programming Based On Similarity

Posted on:2012-09-20Degree:MasterType:Thesis
Country:ChinaCandidate:H W SunFull Text:PDF
GTID:2218330368482081Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
In recent years, there has no better solution to auto-evaluation on C programming language problem, the evaluation of programming problem usually adopts a method that just depends on outputting results to give the grades. This method is simple, but not scientific. It totally ignores the source code of the examinee. However, some automatic evaluation methods based on source code also exist some deficiencies. If we want to get the better effection on the evaluation, we should deal with structure of source code of the examinee deeply.Aiming at the problem above, this thesis proposes an automatic evaluation method based on similarity calculation through the study of the status of domestic and overseas and the related techniques about the automatic evaluation method on C programming language problem, and we analyze and calculate the similarity of the examinees' source code by means of the auto-evaluation framework that is set up with the proposing auto-evaluation method. First of all, the thesis introduces the detailed XML schema definition about the structure of C language to parse the C program. The thesis also designs and realizes the analytical algorithm which converts C program into XML document. At the same time, the conception of program eigenvector is introduced to improve the time efficiency in this paper. The extraction algorithm and the similarity algorithm of the program eigenvector extraction have been designed and realized in this paper. Secondly, we use the test cases to verify the accuracy of the examinee's programs. We join the XML documents and the program eigenvector of the correct examinee's programs and the model answers into the database. According to the scoring points of the answers in the database, we weight the nodes of the XML documents. In this way, we can evaluate the programs more intelligent. Finally, by calculating the similarity of the extracted eigenvector, we find the most similar model answers to the examinee's answers. Moreover, according to the demand of the project, the similarity algorithm of XML documents has been designed and realized. We can obtain the grades of the examinee's programs through the algorithm.The experiment result shows the feasibility and effectiveness of the proposed evaluation method in certain degree, however, there are also some problems to affect the accuracy of the scoring, the paper analyzes the problems and puts forward the corresponding solutions.
Keywords/Search Tags:auto-evaluation, similarity, program eigenvector, schema definition
PDF Full Text Request
Related items