Post date: Dec 21, 2019 7:29:48 AM
SciWrite Tip of a day: Most scholars do not provide original source codes, datasets, and/or clear methodologies in their papers. This practice, often discouraged in academia, limits other scholars to reproduce results (Tables and Figures) necessary for comparisons—a consequence that slows further development and continuity of research, and that encourages (unethical) scholars to falsify their results as an attempt to boost their methods. If you cannot reproduce results, try the following approaches to obtain the resources to compare your method with:
1. Communicate with the authors, usually first (junior) or last (senior), to inquire the resources: source codes, datasets, and experimental settings. Several authors, especially the well-established ones, feel pleased when their works attract interests of the peers. And, therefore, they would collaborate or provide the resources that you can use for comparison.
2. Search the resources in the public repositories:
- MATLAB file exchange (https://in.mathworks.com/matlabcentral/fileexchange/)
- Freesourcecode (http://freesourcecode.net/)
- Sourceforge (https://sourceforge.net/)
- Pantechsolutions (not free; https://www.pantechsolutions.net/)
- Figshare (https://figshare.com/)
3. Ask a question in scholarly social forums, such as ResearchGate, on where you can obtain the source codes/algorithms/implementation flow of a paper that you want to reproduce its results
4. Visit the homepages of the authors, and navigate if their paper forms part of the project and contains the source codes
Do not enter into a trap of falsifying results to boost your method. This academic misconduct can lead to retraction of your work in future, and can mislead other scholars that would like to build their works on top of yours. In addition, try as well to write your paper such that scholars can easily reproduce your results. The methodology section should contain clear thoughts, experimental settings, constants and parameters with known values, and links to your implementation codes.
HIGHLIGHTS: Graphical Abstracts, Reproducing results, Negative Results are Results..!