Sunday, February 22, 2015

The Phantom Degree

The French Minister of Higher Education and Research, Geneviève Fioraso, turns out only to have an English degree and not one in English and Business, as Mediapart (paywall), Le Monde, and  Le Figaro report.

According to Fioraso, somehow "Who's Who in France" mixed up her degrees, and made a double degree in English and in Business out of an English degree with an "option" on Business (screenshot of the entry). She is taking steps to correct this information, she says. Who's Who in France advertises that they verify the degrees from the grand écoles, which does make sense as the Wikipedia entries are free to read, so it would be quite interesting to see how this information came to be in their databases. One must pay 6 € in order to view the entry, however, so I'll stick with the screenshot above.

The government was quick to assure the general public that the minister was chosen for all the great things she has done and not on the basis of a specific degree. Yes, we've heard that before in Germany, in connection with plagiarism cases.

Friday, February 13, 2015

Austrian term papers clog plagiarism detection system

The Austrian online newspaper derStandard.at reports on a bit of problem with their new high school term paper submission system for the school leaving certificates matura. Pupils in Austria are now expected to submit a 40,000 to 60,000 character long term paper (vorwissenschaftliche Arbeit) by the middle of their last year of school. The paper will be graded by teachers and the students must give a presentation on their work.

Of course, since Austria is well aware that there is a plagiarism problem, at least as far as pupils and students are concerned [not so much for doctoral dissertations, but that is another blog post], the term papers must be checked for plagiarism by a so-called plagiarism detection system.

The due date 2015 is Friday, February 13. Surprise, many students have waited until the last minute, and the system is throwing errors that appear to point to the system being swamped. Apparently, they did not also reckon with such large files as are being uploaded. The server operator noted that they were expecting the files to be around 1 MB, instead they were getting 60 MB large files.

Not to fear - there is a Plan B in action: the pupils can submit a printed version at their schools in order to keep the deadline. Or, as one teacher noted in a comment, submit at 5 a.m. The server runs well at that time of the night.

2013 there were almost 44,000 pupils granted their diplomas in Austria. Teachers will now, in addition to grading these papers, have to wade through the results of the plagiarism-detection software, although they also generate false positives as well as false negatives, thus not determining plagiarism but giving some ideas as to where perhaps there could be some plagiarism. Even assuming that a teacher only spends an average of 10 minutes per paper interpreting the results (and this is generous, as the reports are not easy to read and the numbers reported can be quite misleading), this means a minimum of 7-8000 extra hours of work nationwide, but probably tenfold that.

If the pupils are anything like the ones I see in the first semester, they love to take pictures they found on the Internet to spice up their texts - they are much more visually oriented than the older generations. The software will certainly not be able to identify pictures that are not used according to license, so the teachers will also need to use Google's image search or a system such as TinEye to look for the potential sources, increasing the amount of time needed for grading.

Maybe the idea of a term paper submitted centrally needs to be rethought? Of course, they have to learn how to do research and to write about a topic. But we need to be thinking about how to develop methods of assessment that are plagiarism-proof, instead of adding more broken software to a broken system.

Saturday, February 7, 2015

The disappearing links

I've noticed a change in what Google returns when you search on the names of persons who have been documented on the VroniPlag Wiki plagiarism documentation site as having extensive text parallels that could constitute plagiarism in their dissertations and habilitations. It used to be that when one searched for their names, the link to VroniPlag Wiki came at the top of the search.

Now there is a notice that some links have been removed because of personality rights questions, and links show up -- if at all -- past the first page of results. This is perhaps due to a recent court ruling.  The European High Court (EuGH, 13.05.2014 - C-131/12) ruled that people can have links about them personally "forgotten" by search engines. The pages naming them do not have to be removed, but they can insist that the search engines not point to such pages.

So it is not enough to just google a name to see if there are any problems associated with scientific publications about a person. One would now need to know where to look in order to find out if, for example, plagiarism in a doctorate has been documented or a paper withdrawn or issues with a publication documented.

This is unfortunate for scientific purposes, as it is vital that other scientists are made aware of dissertations, papers, and books that have been withdrawn for plagiarism or other academic misconduct. Otherwise they will try and replicate experiments that were forged, or build on top of wrong material. I have heard the excuse that a paper is plagiarized, but the contents are true. That is not always the case, sometimes in plagiarizing, something gets taken out of context and the meaning is changed.

Privacy is important, but scientific papers and dissertations are not part of one's private life. They are contributions to the body of science, and are thus public and open to criticism. That's what keeps us honest as scientists: if we goof up, our names are forever associated with our misdeeds.

Update: There has been some discussion about which cases are affected. The following cases (there may be more) show me the VroniPlag Wiki link either on the second page or not at all: Alm - Bm - Cl - Nig - Rh - Tt