Published 11/04/2016 – Newsletter 67
Similarity detection software is essential to avert the large-scale fraud that occurs when there are no checks. By no means do such programs claim to eradicate the problem by detecting all cases of plagiarism, but they are advancing at the same rate as the tricks used by plagiarists.
The table below summarizes 15 cases examined over the last six months of 2015. It shows the relative usefulness of similarity detection software.
In 50% of cases, the software was inadequate or ineffective.
|Plagiarism detected and proved||A high % of plagiarism detected because it was mostly plagiarized verbatim in English from English to English.|
|Case 2: Law||Plagiarism proved||A high % of plagiarism detected because it was mostly plagiarized verbatim in French.|
|Plagiarism detected||A high % of plagiarism detected because it was mostly plagiarized verbatim in English. However, the large number of sources and the time lapse involved regarding all the disputed texts made further examination very time-consuming even with the software’s help.|
|Case 4: IT||Plagiarism detected||A high % of plagiarism detected because it was mostly plagiarized verbatim. Nevertheless, the software did not detect the plagiarized diagrams and tables.|
|Plagiarism detected||The software detected a % of plagiarism on another source that was incorrectly cited. An examination by an expert in the field discovered other fraud (invented data, false field studies etc.)|
|Case 6: Law||Plagiarism detected||The software detected the sources of plagiarism but an expert in the field had to do the analysis “by hand”. The text was full of paraphrases from several source authors.|
|Two different software programs used. Some plagiarism detected||The software’s result had to be refined: a specialist had to analyze the text to complete the corpus of the plagiarized texts. Some plagiarized passages were separated by several sentences or were not in the same order as in the original article. Nevertheless, it was mostly plagiarized verbatim.|
|Two software programs used. No significant results: very low rate of similarity||Plagiarist method: ambiguous citation to camouflage what was the author’s own text and what came from plagiarized sources. This method combined with using occasional disguises (use of synonyms/paraphrases/ change in verb tense) is hard for software to detect.|
|Inconclusive results||The plagiarism consisted of translations of documents, camouflaged figures and tables that similarity detection software cannot detect.|
|Inconclusive results||Plagiarist’s method: taking others’ ideas and paraphrasing them, but mostly self-plagiarism.|
|Case 11: Anthropology||Software ineffective||Translation and/or paraphrase of a text in French into English: not detectable by software.|
|Software ineffective||The plagiarized books and articles are not accessible on Internet or only via a paying service. In such cases the software gives no result.|
|Case 13: Finance||Software ineffective||Plagiarist’s method: translation of French to English of part of the thesis. Copy/paste of some tables and manipulation or changing results for others. The software is ineffective when plagiarism is by translation or copying tables.|
|Case 14: Anthropology||Software ineffective||Plagiarist’s method: translation from French to English/ plagiarism verbatim and paraphrase. Apart from verbatim plagiarism, these procedures are not detected by similarity detection software.|
|Case 15: Geography||Software ineffective||Plagiarist’s method: manipulation of maps and tables and modification of photos. Software does not take these into account.|