Portal de Periódicos da CAPES

Review of Plagiarism Detection Freeware

2011; Lippincott Williams & Wilkins; Volume: 112; Issue: 3 Linguagem: Inglês

10.1213/ane.0b013e3182096f44

1526-7598

E. Andrew Ochroch,

Academic integrity and plagiarism

In this Issue of Anesthesia &Analgesia, the Editor-in-Chief's editorial “You Will Get Caught” discusses the need to ensure that plagiarism does not continue as a threat to the academic integrity of anesthesiology.1 In his editorial, Dr. Shafer strongly advises authors to screen their own manuscripts before submission to help eliminate plagiarism from the pages of Anesthesia & Analgesia. The availability of software that can scan documents for plagiarism should allow for virtual elimination of the misappropriation of another author's words. These programs break the submitted text into a series of strings that are then fed into a search engine. The length of the strings varies with the search engine. Google will accept a maximum string length of 32 elements (words or separate characters). Yahoo will accept a maximum string length of 24 elements. The major difference between the freely available plagiarism checking programs is the logic of how sentence fragments are produced, the depth of the search, and how the report is formatted. Sentence fragment (string) production is typically done by formal grammatical clause recognition (e.g., commas, semicolons, periods, and, but, or, etc.). However, programs can also randomly produce multiple overlapping fragments to scour the submitted paper for stolen sentences, phrases, and clauses. These strings are typically searched against all documents that have been published or made available through the World Wide Web. Google also searches within Google Books. The search results are then formatted into a report that either immediately appears in the user's browser or is e-mailed to the user. The report will usually list sentences, clauses, or fragments that have been duplicated, along with a link to the original article. Fortunately, much of the functionality of costly plagiarism checkers is now available in freeware. In Table 1, several freely available (November 2010) software options are listed. The freeware typically allows either a “cut and paste” entry of text or a file upload. Usually, the sentences are cut into 8 to 12 element strings. There are several nuances that are worth understanding when using these programs: Reference numbers are often incorrectly included as string elements and, in particular, reduced the effectiveness of the program Crossrefme. Quotations are not always recognized, and as a result, properly quoted text may be classified as plagiarism. Screening an already published manuscript always results in a huge amount of duplicated text, because the paper is referenced to itself. Interestingly, because many web sites will capture and report manuscript abstracts, often there are many copies of the manuscript abstract, resulting in multiple inaccurate “hits” for plagiarism. Pasting text is generally safer than uploading documents. In particular, the internal formatting of PDF documents may insert page numbers, line numbers, headers, and footers into the text. This both hampers the search algorithm and may result in inappropriate matches of the headers and footers. The bibliography should not be entered. If the bibliography is entered, then the program will match the citations in the bibliography to any published manuscript that includes the same citations, artificially inflating the amount of duplicated text. The maintenance and upkeep of freeware is not guaranteed, which should be taken into consideration when a selection is made. There is concern for the potential that copies of the submitted text are maintained by the host site, although many of the programs in Table 1 attest to not keeping copies. None of these programs will be able to determine whether there is a theft of ideas, but they should help to determine whether there has been gross plagiarism. Table 1: Free Plagiarism Detection ProgramsTo test the free plagiarism detection programs, I prepared a document by copying sentences from various sources into a mock plagiarized document. I then rearranged the clauses of the copied sentences as might happen if an author provided minimal alteration to mask the plagiarism. Based on this simple test, Searchenginereports gave the best combination of reliability, speed, and a clear and easy to interpret report. Dustball was a close second. E. Andrew Ochroch, MD, MSCE Department of Anesthesiology University of Pennsylvania Health System Philadelphia, Pennsylvania [email protected]