Initial Trials of Ephorus

This is a Dutch software app which claims to do much the same as TurnItIn (TII), only cheaper. A trial was set up whereby 20 files which had already been passed through TII were put through a demo version of Ephorus.

The results:
• The Ephorus demo account that was being used only checks against the internet. The full TII account used checks against the internet, publications and past submissions.
• In no case did the two systems come up with the same percentage matching text, or the same list of primary sources.
• In 7 out of 20 cases, the matching text scores would lead to the same conclusion by the tutor even though the numbers weren’t the same. In the other 13 cases, the scores were radically different, with Ephorus scores typically much lower than TII.
• In the one case where text from foreign-language websites had been used, Ephorus was demonstrably better than TII (Ephorus scored 62% for the 100% plagiarised text, compared to 31% for TII).
• Whilst Ephorus seems to be better with foreign-language texts, TII has better coverage of English-language publications. This supports Ephorus’ comment that their European-language database is currently much better than their English-language database.

The conclusion is that the use of Ephorus can be justified for checking foreign-language texts, and maybe as a second opinion on tricky TII cases. The reports generated by Ephorus are similar to those produced by TII, available online and as PDF documents.

Operationally:
• Ephorus accepts the same file formats as TII, and takes typically 5 minutes to process a document.
• The Ephorus display shows the matching text, but doesn’t show which bit goes with which source, making it difficult to check the context. It can show alternative sources for the same text, and allow you to skip a source (e.g. an earlier draft), but these facilities are hidden (so infrequent users would forget about them and require re-training).
• File storage is organised on a per-tutor basis, whereas per-class would be better. Not clear how well the user interface would cope with large volumes of coursework. However, these problems go away if the Moodle add-on is used. This adds controls to the standard single-file upload assignment activity so that all submissions are automatically checked and the score made available on the marking page. This is equivalent to the TII Assignment activity.
• If you use the web interface rather than the Moodle (Learn) interface, the Ephorus administrator has to create all tutor accounts, and all of the individual assignment accounts – precisely the scalabilty nightmare that we use the TurnItIn Assignment activity on Learn to avoid. The Learn integration would get around this.
• Ephorus’ adverts are rather naïve in claiming to prevent plagiarism – we are trying to wean people off the idea of ‘plagiarism checking’ and towards ‘text-matching’ because matched text may have been correctly cited, whether or not it is enclosed in quotes.