I’ve noticed that one Document, posted for plagiarism check, returned 100% plagiarism match in ShrePointLMS.
However, when I look the details of this document in Urkund web interface I see 98% plagiarism match.
In most cases plagiarism match percent is the same in detailed Urkund report (Urkund web interface) and shown in ShrePointLMS. But sometimes these values are different.
Urkund service informed us with the following:
"The problem IS on the to-fix list with the developers, though, and will be corrected in time, rest assured. The reason is that the first significance you see in the LMS gets its value directly from the database while the one in the analysis is calculated based on the content on the client side.
As you may have noticed, the significance score in the analysis is dynamic and recalculated based on the actions of the reviewer and the two use different algorithms. If the perception of the score is that it determines the amount of actual plagiarism in a document we do understand that the variation is troublesome. However, the score does not indicate plagiarism. It shows the similarities between the analyzed text and the sources that we index and serves as a “you should probably look at this”-flag paired with a “this much of the document is similar to other sources”-flag. Similarities might very well be deemed perfectly acceptable to a reviewer when they are looked at in detail or may be deemed a false positive, such as a duplicate or an irrelevant find. In that context the gravity of the variation of the similarity score isn’t so severe as it is usually very small and most likely won’t stop a reviewer from opening the analysis itself!"