Text Mining for Plagiarism Detection: Multivariate Pattern Detection for Recognition of Text Similarities

The problem of plagiarism the recent years has been intensified by the availability of information in digital form and the accessibility of the electronic libraries through the Internet. As a result, plagiarism detection has been transformed into a big data analytics problem since the number of digital sources is extravagant and a new document needs to be compared with millions of other existing documents. In this paper, a text mining methodology is proposed that can detect all common patterns between a document and the documents in a reference database. The technique is based on a pattern detection algorithm and the corresponding data structure that enables the algorithm to detect all common patterns. The methodology has been applied in a well-defined dataset providing very promising results identifying difficult cases of plagiarism such as technical disguise.