Plagiarism detection tools check for similarity and not for plagiarism. These tools identify material in your work that matches text from documents in their extensive repositories. When you subject your article for a plagiarism check, the results give a percentage score, known as Similarity Index. The Similarity Index gives an indication of similar contents that has been found in other published materials or theses.

In order to safeguard cases of inadvertent similarities, some limits are prescribed for exclusion. Normally, the following are exempted from similarity checks.

• Information put in quotations with necessary permission or attribution
• References or bibliography, table of contents, preface, and acknowledgements
• Generic terms, laws, standard symbols, phrases, jargon, terminology, and standards equations
• Small matches of common knowledge or coincidental terms (the default value is usually set to 9 or 10. But it differs between countries, for example, in India, it is 14)

Plagiarism detection tools check for similarity and not for plagiarism. These tools identify material in your work that matches text from documents in their extensive repositories. When you subject your article for a plagiarism check, the results give a percentage score, known as Similarity Index. The Similarity Index gives an indication of similar contents that has been found in other published materials or theses.

In order to safeguard cases of inadvertent similarities, some limits are prescribed for exclusion. Normally, the following are exempted from similarity checks.

• Information put in quotations with necessary permission or attribution
• References or bibliography, table of contents, preface, and acknowledgements
• Generic terms, laws, standard symbols, phrases, jargon, terminology, and standards equations
• Small matches of common knowledge or coincidental terms (the default value is usually set to 9 or 10. But it differs between countries, for example, in India, it is 14)

All anti-plagiarism tools help identifying the similarity percentage. It is the responsibility of the author/reserch supervisor to decide whether it is plagiarism or not. Now, the UGC of India has come out with a clear rule that the similarity % should not exceed 10% whereas many universities have their own rule. Some universities go up to 30% similarity as acceptable.

Dear C George Thomas , Terry Purvis

Thank you very much for your kind responses....

The similarity index is an excellent tool to monitor the extend to which some text has been copied/ reproduced without permission/citation and pasted in a research article. However, there have been a number of oceans when some authors have been tracked for such activities. Some software are are made available by the Higher Education Commission in Pakistan to check the similarity index before submitting a PhD thesis, it is mandatory.

In my opinion, it is a good practice and must be adopted by the researchers to eliminate any possibility of plagiarism.

Aslam

Dear Prof. Muhammad Aslam Baig ,

Thank you very much for your kind reply. I completely agree with you. Could you inform us about the the allowable value of similarity index for a PhD thesis or international paper in Pakistan, please? And what can be excluded from this similarity check?

Regards,

Nazar

Nazaruddin Sinaga , Usually the limit is 20% total (which includes citations) by the Higher Education Commission. Like all tools, no index is perfect. What should we do with an experimental section? Experimental section can match in different papers by the same authors. Some people are very good at paraphrasing while copying. This tool will fail. Another level of fraud, especially in social sciences, is using Google translate to steal foreign language works. This similarity index fails. So as readers and reviewers we keep our eyes and minds open to judge the originality of work. No need to panic, try to write on your own and things will be fine.

Dear Dr. M. Farooq Wahab

Thank you very much for your reply and information. In my country it is 25% excluding bibliography. Another one tell us it was 30% in his country. So, for a moment, we can conclude that the allowable similarity is 20-30%.

The big question for us is the meaning of similarity. Most reviewer in my country, and maybe in other countries, think that similarity is a kind of plagiarism. This is what I'd like to make it clear.

Regards,

Nazaruddin Sinaga , You can say that higher similarity index is proportional to the amount of material taken from other sources. So if the reviewers are saying similarity indicates plagiarism, they are right. This is exactly what it is meant to indicate.

• Great discussion! This may give further information! https://www.researchgate.net/post/How_to_find_similarity_index2

This is a very complex question and the answer depends on the field of research. For all papers in the humanities, I would consider that the similarity index should be less than 2%. For papers published in all technical fields, the corresponding upper limit should be 5%. For all cases exceeding these values, the authors should be invited to include a statement at the end of their paper explaining the reason for a high similarity index. The high values of the similarity index may not be intentional.

There is also a question of applicability of existing methods of calculating a similarity index for papers containing many mathematical expressions. The software based on lexicographic search is probably not applicable to such papers.

Dear M. Farooq Wahab ,

Thank you very much for your reply? In your opinion, is the similarity always indicate plagiarism?

Thanks dear Nil Kanatha Pandey for your reply,

The link give us further info about the similarity. Could you suggest us the normal value of similarity index for a submittable paper, please?

Dear Janusz Pudykiewicz ,

Thank you very much for your reply. Please tell us the about the similarity index you mean in your reply. I think it is very difficult to have such the value, cause, normally, if we use around 20 references for our paper, the similarity index just for the references would be 13 - 16 %.....

Dear Nazaruddin Sinaga

Thank you for your important observation. In order to have a truly significant similarity index, we must exclude the references. According to the standard guidelines:

“As a guide, the similarity score, that is the percentage of an assessment which directly quotes other sources, should be no more than 10%. Please note that this refers only to direct quotes within the body of the text and doesnot include the reference list or cover sheets etc.”

For further details please consult:

http://currentstudents.acap.edu.au/assets/ITACAP/IT-Help--Support/Student-Guide-TurnitinFAQs.pdf

We can envision also a test based on the distribution of lexicographic similarities over the different parts of an article. This is particularly meaningful since the high values of the similarity index in different sections are caused by different reasons. Therefore, the distribution of similarities is more meaningful that just a single bulk value for the entire paper. With the emergence of deep learning algorithms, the traditional methods of calculating the similarity index will be improved, being able to detect real similarities beyond those indicated by a simple lexicographic search.

Dear Janusz Pudykiewicz ,

Thank you very much for your reply. After looking to the reference you shared, we would say that the similarity check tools such as Turnitin does not make a decision on what is plagiarised material and what is not. It only measures similarity between the submitted text and other texts.

The important message here is that we can not say that similarity is a plagiarism, right?

Regards.

Dear respected colleagues and scholars. Please participate in the discussion and give your comments.

Thank you very much.

Dear Nazaruddin Sinaga:

As discussed in this group, the similarity index of approximately 10 percent should be fine for a submittable paper, excluding references, acknowledgments and authors' information. But it is not a good idea to copy and paste even a single sentence from the published work. Even if we write in our own words, iThenticate software still may show some similarities between our work and published works, especially in experimental sections and technical terms. I think editors will understand this. As long as we write in our own words with proper credit, it is not a problem even if the software shows some similarities.

Thanks very much,

Nil

Dear Nazaruddin Sinaga

A non-zero value of the similarity index doesn’t indicate intentional plagiarism. The author should consider rewriting the text using different wording to increase the originality of the questionable contribution.

Furthermore, the standard lexicographic search is not sufficient to find papers with plagiarism. We can easily imagine a hypothetical article written using a unique wording but using the ideas of others without providing references. This can happen both intentionally and unintentionally. In order to avoid the latter when claiming novelty, we can always say: “to the authors best knowledge this idea is original....”

The problem of identifying cases of plagiarism and/or copyright infringement is very complicated. In the first step, we start from the standard lexicographic search, this should keep us safe from falling into the trap of using “scientific cliches”. I would recommend performing this check only for the crucial part of the paper where the main weight of the scientific contribution is placed. The second step is much more difficult and often requires a good knowledge of the history of science. A thorough knowledge of the origins of ideas is as important as the first guess provided by the lexicographic similarity index.

Regards.

Thank you very much dear Nil Kanatha Pandey and Janusz Pudykiewicz for your replies...

Dear respected colleagues and scholars. Please participate in the discussion and give your comments.

Thank you very much.

Dear colleagues

Plagiarism is certainly a problem, but hopefully it should not be confounded with multiple discovery. The phenomena of multiple discvovery made in an independant way exists and is real and perhaps more frequent than previously thought in my opinion. Historians and sociologists have remarked on the occurrence, in science, of "multiple independent discovery" for some time. Robert K. Merton (see reference below) defined such "multiple discovery" as instances in which similar discoveries are made by scientists working independently of each other. Merton describes "multiple" as opposed to "singleton" discovery, the latter made by a single scientist or group of scientists working together. Moreover, nowadays because of information and ideas circulating much faster and abundant ways (congress, webinars, open source papers, etc..), the likelihood of plagarism increases but the same applies for multiple independant discovery in my opinion. By analogy, if different pieces of a "scientific puzzle" exist and are known (e.g. through journals), different groups could independently assemble them and reach the same conclusion. Do plagarism checker or similarity index would fail in such cases ?. Many new ideas are the logical ending of already existing ideas. How would one perform plagiarism checker or similarity index if it were applied to the work of Einstein vs Poincare, Maxwell, etc. Nowadays, Nobel Prize laureates tend to be given to different individuals rather than just a single laureate, who often have independently made the same discovery.

Commonly cited examples of multiple independent discovery are the 17th-century independent formulation of calculus by Isaac Newton, Gottfried Wilhelm Leibniz and others, (Hall, 1980).

References:

Merton, Robert K. (1963). Resistance to the Systematic Study of Multiple Discoveries in Science. European Journal of Sociology. 4. pp. 237–282. doi:10.1017/S0003975600000801. ISBN 9780226520704. Reprinted in Merton, Robert K., The Sociology of Science, op. cit., pp. 371–382.

Hall, A. Rupert (1980). Philosophers at War: The Quarrel between Newton and Leibniz. New York, NY: Cambridge University Press. ISBN 978-0521227322.

Dear Alain Robichaud ,

Thank you very much for your complete comment. It make us more clearer...

Dear other scholars, please give your reply.

Thank you very much.

Dear scholars, please join our important discussion at:

https://www.researchgate.net/post/Do_you_agree_if_we_create_an_international_organization_of_university_lecturers#view=5e3cbc1e2ba3a1a3a972a2d2

I think if the similarity index is less than 2% per single source it should not be included in the report. However, this differs from country to country and within institutions of higher learning. I am of the view that the more consensus is required in this regard especially when one is considering publishing in Journals.