Honest similarity reduction means you first find out what each match is, then do the citation and quote corrections, and rewrite the copied parts yourself.
This takes more time than any shortcut. When a thesis comes down from 22% to 8% in this way, the chapters also read better, because mostly the passages that matched were the ones the scholar had not really made his or her own. Many scholars are asking us for a quick method, but no quick method is also safe.
Read the report before you rewrite anything
Most scholars open the report, see a figure like 24%, and start changing sentences from page one itself. Don’t do that. You will lose a week on sentences that were never the problem.
First open the source list. Every checker gives one, and it is sorted by how much of your text each source matches. Write down the first ten sources and the percentage against each. Then ask a few things about every source. Did you cite it? Is it your own paper? Is the match coming in a part that should have been excluded?
Many times two or three sources alone make half the score. One may be a Shodhganga thesis you used heavily. Another may be your own conference paper. The third can be a textbook definition you quoted without quotation marks. The fix is different for each one, and none of them needs you to rewrite the thesis page by page.
After that, check where the matches are sitting. Matches in the review of literature are common, and you can fix them. Matches in your results, discussion or conclusions worry the examiners more, because as per the UGC rules those parts should be fully your own work.
| What the report shows | Usual cause | Honest fix |
|---|---|---|
| A long coloured run of matches in your review chapter | You summarised one source too closely | Close that source, write the part again from your notes, and cite it |
| One short sentence matching exactly, with a citation beside it | It is a quotation, but the quotation marks are missing | Put the marks and the page number, or write the sentence again |
| Your own journal paper comes as a big source | The chapter was taken from your published work | Cite the paper and ask for the exclusion your rules allow |
| Lots of small matches of 5 to 10 words | Common phrases used in your field | Mostly no change is needed. Ask them to apply the small-match filter |
| Matches all over your reference list | The filters were not switched on | Ask for the bibliography to be excluded and the check to be run again |
| A questionnaire or scale in the appendix | Published items you adopted for your study | Cite the scale, and ask whether appendices are checked at all |
Check how the report was run
How fair a report is depends on its settings. Under the UGC 2018 exclusions, some parts should not be counted at all. They are quotations with attribution, the bibliography, the contents pages, the acknowledgements and standard equations. Common phrases of up to 14 consecutive words also should not be counted.
Sometimes your reference list itself comes coloured in the report. That happens only when the filters were not on. Ask library or your supervisor to run it again with quotes excluded, the bibliography excluded and small matches of 14 words or fewer excluded. INFLIBNET itself gives this advice for thesis checks done on DrillBit.
One thesis can give quite different scores on two checkers, and we explain why in our comparison of the three checkers. Before you start rewriting, confirm which checker your university will finally use for the official report, and work on that report only.
Fix quotations and citations first
This step gives the quickest honest drop in the score. Take the source list and the matched sentences, go one by one, and sort each one.
Suppose you copied a definition word for word and you wanted to quote it. Put it inside quotation marks, add the author, year and page number, and go to the next one. Short quotes with the author named are fine as per the rules. For longer ones, you have to set them as an indented block. In APA 7 the limit is 40 words, and other styles count lines instead, so check what your own style says.
Some sentences are summaries where you forgot the citation. Add the citation first. Then read the sentence once again. If it is still following the wording of the source closely, the citation will not be enough, and you have to write that sentence again in your own words from your notes.
One more thing about quoting. If half of your review is inside quotation marks the score will be lower, no doubt, but the examiner will still find it weak, because they see collected text and very little of your own thinking. Quote a definition or a line that is really striking, and summarise everything else. A a reference manager keeps the page numbers safe for you, and our citation generator makes a reference from a DOI in whatever citation style you follow.
Rewrite from your own understanding
For passages that are really too close to the source, the method here works well, especially when English is your second language.
Read the source paragraph two times and close it. Take a blank paper and note down in a few lines what that author found, and what use it has for your own study. These notes can be in Tamil, Malayalam, Telugu or Kannada if that helps you think better. Now from those notes only you write the paragraph in English, and the source should stay closed till you finish.
Now keep both side by side and compare. If yours still has the author’s exact words, change how the sentence is built, and change the words also. You can start with the finding and put the method later. Two sources saying similar things can be joined in one sentence. Then add what this source means for your own objective. Nobody else has written about your objective, so this part will not match with anything.
Synonym changing looks faster, I agree. Say you put “fast” for “rapid” and “research” for “study”. The sentence pattern is still the author’s pattern. Checker will catch the longer runs in that, and when the examiner reads such a sentence he can make out that something is wrong with it.
Your own papers, scales and standard text
A chapter that came from your own journal paper will be matched. If you cite the paper it is not plagiarism, but the number still goes up. Many universities have some way to exclude such papers. Anna University leaves out the papers you claim in the thesis. The University of Hyderabad ignores overlap with papers where you are the first author. Bangalore University wants your supervisor to give a separate certificate that lists exactly which papers can be excluded.
See what your own university rule says. Cite your papers properly in the chapter, and then fill the exclusion form they give. Your published results need not be rewritten only for the checker’s sake.
Some standard material also comes in the report as matches, for example a questionnaire made from published scales, an Act quoted in a law thesis, or a standard equation. Cite all of these. And ask the library whether appendices are checked or not, because in some universities only the chapters go for checking.
Cut what never needed to be there
Sometimes the honest fix is to delete. Look for textbook portions that every thesis in your field keeps repeating. For example, the history of the internet in an e-commerce thesis, three pages on what a Likert scale is, or ten definitions of “job satisfaction” from ten authors.
Your examiner has read these basics in many theses before, and mostly he turns those pages without reading. A shorter literature review chapter that discusses fewer sources and explains how they differ will be stronger, and it will match less also. Method chapters have the same issue. Many are filled with general text about what “descriptive research” means.
Ask whether your examiner would really miss each paragraph, and if not, delete it.
The tricks that fail, and why
Seniors will tell you some shortcuts, and some people are selling them online also.
- Look-alike characters. Some people put a Cyrillic “а” in the place of the English “a”, so that words stop matching. Checkers have started flagging such odd characters. And if somebody copies a line from your PDF and searches for it, the trick is caught then and there.
- Text as images. Here the copied paragraphs go into the thesis as screenshots. The report cannot read text inside an image, that part is correct. But one look at photos of paragraphs in a thesis, and the examiner is going to ask you the reason.
- White text and extra spaces. Some hide copied words by making the font white, and some others type a space after every letter. Many libraries convert your file to plain text, and this shows up at once.
- AI paraphrasing tools. These change the words, and many times the meaning too. INFLIBNET’s DrillBit service gives an AI report with every submission, and Osmania caps AI-generated content at 20%. A paraphrased thesis may clear the similarity check and still fail in the AI check.
If the trick is found out later, you face a bigger problem than a high score. Then it is no longer about revising a chapter, it becomes a misconduct inquiry against you. The UGC rules also cover plagiarism that comes to light after the award, and in that case the degree may be held in abeyance.
Plan your checks around the limits on attempts
Most universities allow only a few attempts at the official check. Under its PhD Manual, VTU allows three submissions in all, with fees of ₹5,000 and ₹10,000 for the second and third. JNTUH gives three chances, and each revision has to come within a month. At Bharathiar the certificate is valid for 15 days only.
Do your own cleaning on a draft first, with time in hand. Use the official check for the final version only. If you are using a ShodhShuddhi account, INFLIBNET does not save your uploads to its repository, so your revised file is not matched with your own earlier upload.
If you want help to read the report, our similarity check and reduction service goes through the source list along with you. We mark which matches need a citation, which need quotation marks and which need rewriting in your own words. The rewriting you will do yourself. In the end of it we check that the meaning is still yours and that the score came down for the right reasons.
Questions scholars ask
Can I reduce similarity just by adding citations?
No, citation alone will not do it. Citation only shows the reader whose idea it is. The checker does not look at your citation at all. It looks at the words, and if they are copied it shows them in the report. Along with the citation, either put the exact words in quotation marks or rewrite the sentence yourself.
Is it wrong to use a paraphrasing tool for a few sentences?
It is risky, even for a few sentences. Many times these tools change the technical meaning, and some checkers now give an AI report next to the similarity report. Better to rewrite those sentences from your own notes.
My report shows 30% but most of it is my own published paper. What now?
First cite the paper inside the chapter. Then check what your university says about self-citation. Many universities exclude your own papers if you submit a form or a certificate from your supervisor, and after that you can ask for the report to be rerun with that exclusion.
How long does honest reduction take?
That depends on where the matches are. If it is only citations and filter settings, one or two days is mostly enough. If a review chapter has to be rewritten from your notes, give it one to three weeks. Start well before the official check, not after it.
Will translating a source from another language avoid the match?
The checker may miss it. Still it is copying, if you show another person’s text as your own. Use the idea if you need it, with a citation to the original. But the sentences should be written by you.