Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

Version 1 Next »

Scientific writing is no miracle. It follows certain standards and structures which are similar in all areas of publication. Especially with papers being short in length - some conference-papers are supposed to not exceed maybe 4 to 8 pages - the structure and precision of working is keen to success. This page is written by Thorsten A. Kernto summarize experience of 20 years of publication in engineering context, and having reviewed douzends of papers.

The basic but simple target of each scientific publication is, that if you read it and do now know much about the topic you should be fascinated, you should understand the challenge, you should learn something which you can apply if you step into the topic.  Being an outside you should love to read your paper. If only as an insider your like it (because the topic was so fascinating and you are so proud of your work), this is not enough value for a publication. Your paper must allow the scientific community to learn, it is not (only) a documentation.

The following structure shall give you some guidance:


Lead QuestionContent expected
What is the challenge?Describe the engineering challenge cleary derived from a need, e.g. of society, of other engineering challenges, of prior work 
What is the base-hypotheses of your work - the scientific question?Based on the challenge you usually define a scientific question your work follows. Weak questions are something like "to generally improve control" or "to find a solution for", strong questions are "to improve control to lower than 5% control error" or "to find a solution reaching the bandwith from 100 Hz at 5 N amplitude". In other words, the scientific question is usually quantified. Not necessarily in the same sentence, but this is the "requirements"-section based on which you will later evaluate the performance of the solution
What is prior work, how did others solve it?There is always prior work. It may not be precisely your question, but it always is related. In this part the top 10 related topics are cited. Make sure to catch the most important ones. Good guidelines are always to check for the number of citations of the source you choose and try to find the "top-cited-author" in the area of your research. Usually you are searching for two extrems, the eldest fundamental paper most people refer to, and the newest publications who cite the top-cited author.
Your solution, the fun partIn this part you are describing your solution. As we usually are discussing multi-domain-topics in mechatronics, this part frequently consists of a mechanic, a control and sometimes an electronic component. Try to avoid trivial topics. There is NO VALUE in Arduino or Rasperry circuits. Although this may be something you found extremly interesting and a big learning for yourself, believe me, it is not for engineers in general and especially not for the reviewers you are going to face in a peer-review. You should mention the components you used as precisely as possible, but stick to unusual things, not to things you can find with 1h search in the internet!
Verification according to hypothesisThe next step is to verify your design according to the initial hypothesis. Make sure to focus on answering the questions and requirements from your hypothesis, but on top of that extend your findings by some additional discussions. Show data, show them in a physical base (time, frequency, m/s2), do not show based on "samples taken" or "discretization". For verification, you should have spent equally as much time as for the design itself. The experiments must be chosen carefully, the material must be prepared in high quality and the measurements must be beyond any doubt. So if you did not plan your verification and your experiments are done last-minute, do not publish!  In HMIs verification frequently needs the involvement of people and the execution of psychophysiological experiments. If you do this, make sure that you are following the concept of psychophysical studies and argue in the standards (JND, DL, Confusion-Matrix, ...) which are typical to this domain. Do not just say: "People liked it", no one will believe you!
OutlookThere is an expectation that you create/discuss two directions in your outlook. One part must be data-driven and is more or less a "next-steps" kind of thing based on the data availabe from your experiments. The other part can be highly speculative. Ideally it combines your research with the prioir-art research and formulates a next hypothesis.
Final sentenceDo not forget to mention grants, if you had some which were financing you. 


AbstractThe abstract is of course the first part of the publication. But honestly, the abstract is written at the very end. It is like a short-version of the paper. It must be conclusive, touch all topics above (one sentence per topic).


  • No labels