Plagiátorství v textových dokumentech: Metody odhalování plagiátů

This thesis is devoted to detection of plagiarism among documents in large document databases. The problem of detection of plagiarism is more appealing today than ever. Easy accessibility of documents in digital form contributes to this problem. To enforce author rights and wipe out plagiarism it is necessary to project such system that will be able to distinguish plagiarism among documents with certainty. Such system is valuable help in academic field, where it can be used for controlling of student's final thesis and seminary works. Main attention is devoted to searching of plagiarism in Czech and Slovak language. Basic definitions of plagiarism are defined in the first part of this thesis as well as basic problems, which this system must address. Systems for detection of plagiarism are classified in this thesis and their individual components are carefully analyzed. Theoretical analysis of these components is followed by experimentally gained information and recommendations. Main attention is devoted to extraction of text information from the documents, document standardisation and search for best fitting inner form of documents. In the last chapter methods and algorithms, which can be used for calculating of match index between two documents, are discussed and compared with each other. Considerable effort is devoted to system performance. The main contribution of this thesis is the comparison of usage of word bigrams with longer n-grams. Also the possible contribution of information about sentence division of the document is analysed and presented.