Posted on : May 19th 2023
Posted by : Srinivasan Govindarajan, SVP Strategy & Solutions, and Shanmugapriya Kandavelu, Deputy Manager
As discussed in Part 1, while similarity check reports are an important tool for ensuring research integrity, they also have their limitations. In Part 2, we will explore how to navigate these limitations and maximize the benefits of similarity check reports through best practices for accurate results.
Although similarity check reports are a useful tool for checking the originality of a document, they have significant limits and may produce false positives and false negatives.
False positives occur when the report identifies matches with other sources, but these matches are not indicative of plagiarism. This can happen when common phrases or references are used in the text. For example, a manuscript is submitted to a journal and runs through a similarity check, which returns a high similarity percentage. However, upon closer examination, the editor realizes that the manuscript is an extension of the author's own thesis or preprint, which has already been published online. Since the author's own work is the source of the similarity, there is no potential for plagiarism or unoriginal content, and the high similarity percentage is a false positive. False negatives, on the other hand, occur when the report fails to identify instances of plagiarism in the text. This can happen when the text has been paraphrased or reworded to avoid detection. For instance, after a manuscript undergoes a similarity check, the report indicates a low similarity percentage. However, the editor notices upon further inspection that the manuscript's title is an exact match to a previously published work and that some figures, tables, and numerical values are also similar. Despite the low overall similarity percentage, the editor identifies a potential concern for plagiarism or unoriginal content, rendering this a false negative.
Another limitation of similarity check reports is that they only compare the text with existing documents in their database. They cannot detect instances of plagiarism in unpublished works or works that are not available in the database.
To avoid false positives and false negatives, it is essential to thoroughly evaluate each match and determine its relevance to the content. In addition, it is strongly recommended that several similarity check tools be used in order to guarantee the report's precision.
The CrossRef Similarity Check service helps publishers and editors to detect potential plagiarism in submitted manuscripts. The Similarity Check service is powered by iThenticate, a plagiarism detection software that uses advanced algorithms to compare submitted manuscripts against a vast database of scholarly literature, as well as web pages, books, and other sources. When a manuscript is submitted to a journal or publisher, the iThenticate software generates a similarity report that highlights any potential matches between the submitted manuscript and other sources. Additionally, iThenticate compares submissions with any existing documents including Crossref posted content, and can also detect instances of self-plagiarism, where the author reuses their own previously published work.
Similarly, for CrossRef Similarity Check use, publishers also need to provide their content in a format that is compatible with Crossref's database. This enables the Crossref database to identify potential matches between the submitted manuscript and other published sources, making it easier for publishers to check for potential plagiarism and for researchers to access related academic content sources.
Using similarity check reports alone to detect plagiarism has its limitations. To overcome these limitations, it is crucial to supplement the reports with additional tools, such as expert review. Furthermore, educating authors on appropriate citation practices and the consequences of plagiarism can help prevent it from happening in the first place.
While similarity check reports are crucial in identifying potential instances of plagiarism or duplication, the interpretation of the results is equally important in making informed decisions. The score alone cannot determine whether a manuscript is compromised or not, as there are often legitimate reasons for high similarity scores, such as the use of common phrases or terminology within a field.
Therefore, a combination of technology, analytics, and expert review is necessary to ensure an optimal balance between identifying compromised manuscripts and not discarding good manuscripts. This can involve a thorough examination of the context and source of the matched content, as well as an assessment of the author's prior publications and research history.
Additionally, a human review can help to identify instances of accidental or unintentional plagiarism, which may not be detected by automated similarity checkers alone. Therefore, a collaborative approach between technology and human expertise is essential in ensuring the integrity of academic and scientific publishing.
Similarity check reports can be a valuable tool for editors in ensuring originality and preventing plagiarism. However, it is important to follow best practices to maximize the benefits of similarity check reports and achieve accurate results. Following are some best practices to follow when using similarity check reports
Properly prepare the document: Ensure that the document being analyzed is in the correct file format and has consistent formatting throughout the text. This can help the similarity checker produce more accurate results.
Choose an appropriate similarity checker: Different checkers have different algorithms and databases, so choose one that is appropriate for the type of text being analyzed.
Review the results carefully: Don't rely solely on the percentage of similarity provided by the checker. Instead, review each highlighted section of the text, comparing it to the original source material to determine whether it constitutes plagiarism. Keep in mind that some highlighted text may be properly cited or paraphrased and should not be flagged as plagiarism.
Work with the author to address issues: If plagiarism is identified, work with the author to address the issue. Depending on the severity, this may involve suggesting revisions or even rejecting the work outright. In some cases, it may be necessary to report the plagiarism to the appropriate authorities.
By following these practices, editors can ensure that they are providing accurate and high-quality content to their readers, all while maintaining integrity and ethical standards.
Similarity check reports play a vital role in upholding academic integrity by finding similarities between submitted works and existing sources. They help to detect and deter plagiarism, provide valuable feedback to authors, and highlight potential areas for improvement in their research skills. Different types of similarity check reports are available, and they may vary depending on the software or tool used to generate them. However, it is essential to understand the limitations of these reports, including false positives and false negatives, as well as their inability to detect plagiarism in unpublished works. Therefore, it is crucial to manually review the report to ensure that any flagged instances of similarity are actually problematic. It is recommended that similarity check reports be used as a guide and that each match be reviewed carefully to determine whether it constitutes plagiarism or not.
Advanced algorithms for identifying instances of plagiarism in paraphrased or reworded language, as well as the incorporation of AI to enhance the reliability of similarity check findings, are potential areas for future research.
R. C. (n.d.). Similarity Check - Crossref. www.crossref.org. https://tinyurl.com/4ppbt9wy
H. (n.d.). Interpreting similarity check reports. Hindawi. https://tinyurl.com/2p98687y
Interpreting the Similarity Report. (n.d.). Interpreting the Similarity Report. https://tinyurl.com/yjff4x4b
Working with an originality report: false positive and false negative results. (2018, December 19). Check Plagiarism Blog - PlagiarismCheck.org. https://tinyurl.com/474raupf
The availability of research data is essential for ensuring the reproducibility of scientific findings. In recent years, publisher’s submission requirements have encouraged data sharing to improve the transparency and quality of research reporting. Data sharing statements are now standard practice.
Change is a heterogeneous disruption, and digital transformation is no different. It is inevitable to business today as change is to life, but how companies employ it to orient technology for the larger vision of their business makes all the difference.
Peer review is in high demand, despite its inherent flaws, which range from the possibility of bias among peer reviewers to procedural integrity to the stretch of time to publication.
Two new forms of peer review have emerged in the last two decades - post-publication peer review, in which manuscripts are evaluated after publication; and registered reports, in which publications are examined prior to submission to the journal
The push for Open Access publication has been around for more than 30 years now. The past year and a half, however, has produced an exceptional case study on the potential of Open Access.
Our solutioning team is eager to know about your challenge and how we can help.