5 July 2018

Research Pick: Did you mean to say that? - "Plagiarism detection based on semantic analysis"

Plagiarism, wherein an author or other creator, simply copies the original output of another and passes the work of as their own with giving due credit is on the rise, it seems, particularly in the realm of research. An author can simply copy and paste great tracts of text from another author and hope that the editor who receives their manuscript is disinterested in checking that the submitted text is entirely original or otherwise lacks the skills or inclination to check.

Most plagiarism detection software which seeks to flag such offences compares chunks of text in a larger document with documents in a database or searchable on the web. There has been little attempt to look at context and semantics. This is an ongoing problem as a plagiarist may copy and paste whole paragraphs and pages, they may also be wily enough to change some of the text order or substitute synonyms in their version of the plagiarised text for the purposes of obfuscation.
Researchers in India suggest that their semantic analysis of text reveals similarities and so could lay bare that kind of fraud.

Mukherjee, I., Kumar, B., Singh, S. and Sharma, K. (2018) ‘Plagiarism detection based on semantic analysis’, Int. J. Knowledge and Learning, Vol. 12, No. 3, pp.242–254.

No comments: