On the Number of Search Queries Required for Internet Plagiarism Detection

Abstract

Digital plagiarism detection tools have become an integral part of learning management systems in recent years. Discovering Internet plagiarism in student papers has attracted a lot of attention from both researchers and software developers since the beginning of the WWW phenomena. In this paper we discuss how the time performance of such search systems can be improved by reducing the number of search queries. The study shows that if a significant part of a paper was plagiarized from the publically available on-line sources, then only a relatively small number of search queries are required to reliably locate the possible source for detailed comparison and manual confirmation of plagiarism. Indeed, we show that a 5% selection is enough to find the source of plagiarized paper and using more than 40% of the text in the search queries does not improve the search results. These findings help us to recommend the default search parameters for a plagiarism detection plug-in for open source course management systems reduces the time required for querying the search engine thus decreasing the workload for the plagiarism detection system.