Plagiat.pl is a Polish company with 20 years of experience, whose clients include almost 200 publishing houses and 700 universities in Poland and around the world. The company's Antiplagiarism system, which is available to Lodz University of Technology Publishing House, is a helpful tool for authors and editors of books and articles. They receive a full report on the analyzed text, so they can correct any errors and shortcomings before sending the text for publication. The antiplagiarism report is not intended to stigmatize anyone, but only to point out spots which, due to an oversight or incorrect marking (missing quotation marks or distinguishing a quotation in the text), are detected and highlighted by the system as potential inappropriate borrowing. In collaboration with the authors, all instances of such unintended failings are eliminated.
Artificial intelligence or else who is the author?
As of August 2023, the Antiplagiarism system has been enhanced with an AI detection module, which detects the likelihood of a text being written by artificial intelligence. As a result of the analysis, a detection report is produced, giving the probability of a text being produced by GPT-based tools as a percentage. GPT – Generative Pre-trained Transformer, is GPT is a large machine language model based on huge collections of texts. It is capable of generating autonomous texts according to the context fed to it. The AI detection module also makes it possible to identify those parts of the publication where artificial intelligence may have been used. The new tools used in editorial processes serve primarily to guarantee originality and research integrity, thus ensuring a higher quality of future publications.
The verification process in the Antiplagiarism System
A manuscript that is received by the Lodz University of Technology Publishing House for antiplagiarism check (acceptable formats are doc, docx, pdf, odt, rtf, ppt, pptx) is prepared for scanning in the Antiplagiarism system. Eliminated are elements such as affiliations, abstracts, graphic materials, tables, bibliographies or abstracts. Only the content of the text is kept, and it is compared against the following text databases: RefBooks, the Database of Legal Acts, the Lodz University of Technology's home database, the Database Exchange Program and - most importantly - the Internet resources. Next, a MS Word file is uploaded into the Antiplagiarism system and the verification procedure is launched.
Depending on the length and complexity of the text, the procedure may take anywhere from a few minutes to an hour or so. Once it has been completed, the publication is marked as verified in the Antiplagiarism system, and the editor receives an e-mail with three similarity reports: summary, full, and interactive ones. The reports, along with their interpretation, are sent by the editor to the author(s) of the manuscript.
Summary Similarity Report
This PDF report contains only statistics, with no specific excerpts from the text. The following items are included:
• Alerts - editing interventions possibly intended to distort the results of the analysis, e.g., insertion of characters from a different alphabet, bolding, microspacing, hidden characters, paraphrasing.
• Similarity metrics - here three coefficients are examined: Similarity coefficient 1 (WP1) - should not exceed 50%; Similarity coefficient 2 (WP2) - if larger than 5%, the proofreader is required to perform a detailed analysis; BAP coefficient - shows what percentage of the work is sourced from the Wolters Kluwer Poland Legal Act Database. If the standard values of the coefficients are exceeded, the author of the examined work is provided with an explanation of what the problem might be and how to fix it.
• Active similarity lists - presented here are the ten longest fragments of the detected borrowings and individual covert citations color-coded according to their source. Among the most frequently occurring in the analyzed works are fragments cited from the Internet, and they are marked in green. Provided here are the URLs from where the respective borrowings originate.
Full Similarity Report
A full similarity report provides all the data and information that the summary similarity report does plus the content of the analyzed work, with the questionable fragments, i.e. borrowed material, marked in color. They require reviewing to ensure that they are properly marked with quotation marks or referenced in a footnote.
Interactive Similarity Report
An interactive similarity provides all the data and information that the full similarity report does. Additionally, the data and information are displayed on a split screen In addition, the data is presented on a split screen, which shows statistical data on the left ( just as in the summary similarity report), and on the right the body of the text with the highlighted passages. On the left you can open a web page where the passage you are looking for is highlighted, whereas on the right you can find the relevant paragraph in the manuscript under analysis. The interactive report allows users to compare the text that is being verified for originality with the material that might have been plagirized.
How to make sense of the interactive similarity report
In the text displayed on the right side of the screen passages that are similar to those in other sources as well as quotations are highlighted. Quotations that are properly marked with quotation marks are lilac-coded. Coded in green are the fragments that the system matched with fragments from Internet sources (in any language). Coded in red are fragments identified in the home database of the Lodz University of Technology and the Database Exchange Program. Highlighted in orange are excerpts found in the RefBooks database (a database of multimillion scientific texts in Polish and English from all fields of science, includes the resources from Paperity, Termedia, CH Beck, ArXiv, and many others), while the blue background color is used to distinguish excerpts from the Wolters Kluwer SA Database of Legal Acts. The two shades of green and red are used to distinguish separate excerpts (either from different sources or from different parts of the same source). The system also highlights (underlining, paler shade) phrases that have undergone minor modifications, e.g., word order changes, use of closely-related words, addition or deletion of words. The original phrase can be retrieved by highlighting the modified passage.
Displayed on the left side of the screen is all the data and information that the summary similarity report provides. The "Alerts" box allows you to track the number of times a borrowing occurs and to navigate to the points in the text where it occurs. The detected sources of similarity are shown in sorted lists. The first of these lists contains the ten longest passages that have been identified in other sources. The list shows the source data, similarity statistics, and a button to be taken to the exact spot in the text. Spots in the text that are marked in bold indicate that you are dealing with fragments of 25 words or more (WP2).
For passages that converge with the Internet sources you can view the web page on the left, the text on the right, and navigate between the two. The verifier can add comments to the text. The interactive report allows you to add an evaluation directly in the system.
At the Lodz University of Technology Publishing House, the results of the originality verification (interpretation) are provided in the content of an e-mail with attached similarity reports, summary and full, and, at the request of the authors, also the interactive report.
Home page: https://plagiat.pl/