Análise e otimização de desempenho da ferramenta Miss Marple: detector de plágio

Abstract

The research area of automatic plagiarism detection of digital papers is still in development. The methods and studies about this subject these days are new and small amount compared to other areas of computer science, while the plagiarism practice in academic papers is more and more easy due the increasing amount of information available on the internet every year. Within this scenario is Miss Marple Plagiarism Evidence Detection Tool Based on PED Method Plagiarism Evidence Detector. Such tool performs detection of plagiarism evidence in portuguese texts through web searches for suspect documents and then compared in terms of content similarity. However, it showed poor performance during its execution due to a unsatisfactorily long processing time compared to other similar tools. Before this situation and after conducting bibliographical studies in plagiarism detection area, it was found that it would be possible to restructure its code and optimize its performance, as well as implement a bilingual plagiarism detection feature, which it did not have yet. After the necessary changes were, a set of runtime tests was performed with selected documents in order to validate the updated version of the tool. The test results proved that the changes that have been made improved the performance of the tool compared to its prior version as well as introduced the bilingual plagiarism detection support. This work shows how the development process of the new version was conducted, since the identification of the initial issues, the bibliographic study, description of the changes that were made, the tests and their results.