Working plagiarism tools

This text is based on the article in the newspaper "El País" published on September 14, 2018.

The link to the original article is as follows.

How do these programs work?
Users upload the documents they want to compare into the tool.
Sometimes the solution is integrated into the virtual platforms in which students load their work, such as Moodle or Blackboard.
At the Universitat de València you can find our virtual classroom platform (Moodle)
These media directly analyze the documents before passing them to the teacher.

How do the results look and what do the percentages of possible plagiarism mean?
The tool searches the text strings and compares them with the documents in their databases.
The user receives as a result the document he has compared and, if there is a match, the matching fragments are marked in another color.
The final percentage is the percentage of all the text that appears in another document.

From what percentage is plagiarism?
It depends on the field of study, says PlagScan managing director Markus Goldbaech. "A 20% match in fields like the right,
where it is usual to quote literally long fragments of jurisprudence, it should not be strange ".
But in fields such as the natural sciences, above 3% "may be indicative of a possible copy".

"It's hard to give a percentage", "[The percentage] is very relative. The program gives you a global percentage and percentages of each site that finds content.
You may have picked up 10% of one site, 15% of another. If it gives you a tremendous sum you have a problem,
but you would also have it if one of those texts has plagiarized 50% or more. It is valued considering both things. "

What about plagiarizing another language? Here the tools run into a huge difficulty, they say from PlagScan.
"The problem has not been solved yet, and the match is not perfect in any language."
The problem is that very different translations can be made from the same source text.
One of the possible solutions to overcome this problem is to compare not only literal strings of text, but also coincidence of ideas, although they are not expressed literally the same.

What can escape? No company has 100% access to scientific and academic production.
In addition, not all contents are fully digitized or with their texts in comparable format.
PlagScan believes that plagiarism is detected 70% of the time. The remaining 30% escapes.
What other aspects, besides the coincidence, alert of a possible plagiarism?
Analysis algorithms also value the way an author expresses himself (for example, the most usual grammatical structures he uses).
If a paragraph leaves the pattern in the same text, it alerts to possible plagiarism.
But we must remember that paraphrasing is common in academic texts, and does not imply plagiarism.
"What is the difference between a translation and a paraphrase? It is very difficult to determine," asks Markus Goldbaech.

» FAQ de Aula Virtual