Problem with V895

[Stella.Kritsotaki]

I wrote to your support team

"I bought  "Plagiarism-detector" on January 4th of 2014 as you can see on the attached mails I received. Until now I was very happy with the product. I have a small business that prepares academic essays and helps students prepare their own original essays. From the 5th  November when I downloaded your last file "pd_setup" I am facing problems. The reason is that the programs distinguishes between arts and sciences and has two options of comparing essays. The science version detects only "word to word" exactness while the arts version is very sensitive in paraphrasing the same answer and shows very high plagiarism percentages. These high percentages of the Arts version are not reliable because the answers of the students are the same – because the questions are the same - but written with completely different words and in completely different style. So these 2 separate versions are not suitable for me because the science is much less sensitive and the arts much more sensitive than  appropriate to let's say check a marketing essay that students are obliged to answer the same questions so they have to say the same things more or less. The problem would be solved if I was able to work with the previous version however the program leaves me no option to avoid downloading the new one.
I see these solutions: Solution1 to be able to work with the previous version and avoid downloading the new version
Solution 2: Decrease the sensitivity of the arts version so as to reflect the sensitivity of the old version.
I am waiting to hearing from you soon because these last days I am doing only manual plagiarism and it is very tiring and time consuming."

The answer is:
Thank you for the detailed feedback.
We will consider decreasing detection rates for the Arts, as this is not the first request like this.

As for providing the previous version - unfortunately that is not an option. Due to some third party changes, older version now have a very poor detection rate. That is why we have enabled force update to the current version.

My new question: Since until then the program is useless to me when can i expect such a decreased detection or at least the ability to manually adjust it. You understand that if nothing happends within let's say a week i will switch to another program and would definatelly not recommend yours. Mistakes should be corrected!

[Alexei B.]

Like you were informed, we will consider the detection threshold for the Arts. However as of now we don't see enough reason to change it - this preset is aimed at detecting obfuscated plagiarism and our tests showed it deals with this problem just fine.

We need enough evidence of false-positives with this preset to change it. We are ready to analyze each such case and we have some of them stored already. But the number of such cases is not yet enough to start such changes.

We recommend to use word-to-word preset when it is necessary to avoid false-positives.

[Stella.Kritsotaki]

I can understand that in order to change the "arts" version you need to receive several negative feedbacks. However i was informed through e-mail that many other clients also reported very high plagiarism rates in essays that were different in style and expression but had similar content because the subject was the same. Another recommendation would be to unlock the ability for a customer to use the previous version of the software if he finds it better in comparing results at his own institute. I understand you consider it obsolete however it worked better for many clients. If your company does not want to take responsibility for any plagiarism misdetections just pop up a window telling the client that "unless you have installed the newer version you take full responsibility for any false results... etc". It is important to give clients degrees of freedom and keep them happy. For me unfortunately if a new version does n' t come out soon or if i am not able to return to the previous version i will have to switch to another program. Your suggestion about word-word exactness does not solve my problem because it is very weak in detecting plagiarism even in scientific essays and therefore is no option for me.

[Alexei B.]

Another recommendation would be to unlock the ability for a customer to use the previous version of the software

This would have been done, if it were possible. It was not our wish to force updating to the latest version, but the need. Due to some third party changes, versions 850 and bellow started providing totally inadequate results (very poor detection), thus we were to force-update our clients.

T-comparator presets are not so easily changed. Presets currently used are based on the algorithms developed for PAN'14 conference, plagiarism detection task, and showed one of the best results with the used document corpus. One of the lines of our research now is what is different between that corpus and the "real-life" cases, which can be closely connected to the problem you report.

Changing the preset in the current system requires some research, as we need to be sure the new preset works better. At present we have started additional tests on the false-positive detection and upon the results we will both post here and may start making the needed alterations.

[Zodiass]

I have problems like this.