Text similarity measures and plagiarism are quite different terms. Using paraphrasing you can reach 0% similarity index while still plagiarize. On the other hand you can write contributing paper that to introduce to the domain problems you are solving, has to reference a lot by citing others researchers.
Text similarity measures and plagiarism are quite different terms. Using paraphrasing you can reach 0% similarity index while still plagiarize. On the other hand you can write contributing paper that to introduce to the domain problems you are solving, has to reference a lot by citing others researchers.
It is also worth noting here that different plagiarism detection platforms will give different results for a given paper, so talking about % index without focusing on concrete platform and concrete case is quite pointless.
Checking one's paper on popular and free platform and gaining satisfactory result will not guarantee that this result will be confirmed during checking this paper while it is submited to a journal because usually editorial boards use more realiable plagiarism detection plaftoms.
This is because effective detection plagiarism requires primarily the current and complete database of scientific texts in a given field. That's why it's hard to talk about the program, but rather about the (web) service. There are several such services, but most of them are paid (see, for example, https://www.crossref.org/fees/#similarity-check-fees). Gathering full texts of scientific articles (and not just searching the resources of the Internet) must generate costs, so effective detection of plagiarisms can not be costless too.
It has no fixed answer. As most journals and research institutions have set different standards so it is better to check the journal's info etc. where are you submitting your work. Furthermore 10% similarity index may be an out-of-the-danger percentage.