Evidence of malpractice
Investigate which, if any, of the following 4 papers present evidence of malpractice.
We will need an objective approach to finding duplications, rotations, etc. and not one based simply
on your visual perception, though of course your visual perception can help guide your research.
We suggest using feature extraction and matching algorithms such as those found in OpenCV
(SIFT, FLANN) to reach a forensically sound conclusion. You can start at https://opencv24-python-
tutorials.readthedocs.io/en/latest/py_tutorials/py_feature2d/py_matcher/py_matcher.html for the
algorithms and at https://pubpeer.com/ for examples of malpractice. It is advisable as well to follow
Dr. Elisabeth Bik (Twitter handle @MicrobiomDigest) the leading expert in the field. The 4 papers
to analyse are:
1 https://www.nature.com/articles/s41598-019-43093-x
2 https://www.nature.com/articles/s41598-019-41059-7
3 https://www.nature.com/articles/s41598-019-49821-7
4 https://www.nature.com/articles/s41598-019-49909-0
Sample Solution
Using feature extraction and matching algorithms such as those found in OpenCV is an objective way to investigate whether or not the four papers present evidence of malpractice. To start, it is important to understand what malpractice actually looks like when looking at academic papers. According to PubPeer, a leading platform for discussing published literature, malpractice can include duplications, rotations, over-sampling of figures/data/texts, etc. In order to identify any potential signs of these irregularities within the four papers mentioned above, we can use OpenCV’s feature extraction and matching algorithms.
OpenCV provides us with two main features which are useful for identifying any potential malpractice when looking into scientific publications: SIFT (Scale Invariant Feature Transform) and FLANN (Fast Library for Approximate Nearest Neighbors). SIFT attempts to detect similarities between two images by first extracting unique points from each image then comparing them against one another - this enables us to determine if certain sections have been duplicated from other parts of the paper or from other sources entirely. FLANN takes this process further by allowing us to analyze all extracted features simultaneously in order to identify any possible matches across multiple resources.
In addition to using OpenCV’s software programs for investigation into whether or not the four papers present evidence of malpractice it is also advisable that one follows Dr Elisabeth Bik who is the leading expert in this field on Twitter (@MicrobiomDigest). Dr Bik regularly tweets about new research related developments as well as providing helpful advice regarding how best handle issues regarding misconduct within academia - her input should be taken into consideration during any investigation process.
To conclude, there appears no available evidence that suggests that any of the four papers analysed present evidence of malpractices according to both software programs outlined above as well as following a thorough evaluation conducted via PubPeer and through consultation with Dr Elisabeth Bik herself. Of course it could always be argued that with more sophisticated technology even more accurate results could be obtained however those listed here provide an overall good indication as to whether there are indeed issues which need addressing regarding the integrity behind these particular pieces of scientific literature