SSTI: Semantic Similarity to detect Novelty of Thesis Ideas

Abstract

The act of plagiarism in a thesis can decrease the quality of a student’s thesis. The plagiarism is perhaps unintentional. One form of this plagiarism in the thesis is the similarity of ideas in terms of topics, methods and cases used. Detecting the similarity of ideas has its difficulties because it must be able to understand the context of a document. This research implements a similar idea detection framework with a semantic similarity approach. Calculating the similarity value between the thesis proposal and the crawled data is carried out by considering the cluster distance and the hierarchical structure in the knowledge base. Comparative data were obtained from open portal publications. The evaluation results return accuracy is over 100 data testing.