What to Read
Literature Review
- Ten Simple Rules for Writing a Literature Review, Philip E. Bourne, 2013, https://dx.doi.org/10.1371%2Fjournal.pcbi.1003149
- High level tips for writing a literature review.
- How to Write a Literature Review Paper?, B. V. Wee and D. Banister, Transport Reviews, vol. 36, no. 2, pp. 278–288, Mar. 2016, doi: 10.1080/01441647.2015.1065456.
- You can take a class from Bert van Wee on this process.
- Engineering: The Literature Review Process https://libguides.asu.edu/engineeringlitreview/start
- A guide from Arizona State University on writing and engineering literature review.
Thesis/Dissertation
- Writing a Scientific-Style Thesis: A Guide for Graduate Research Students, NUI Galway, Dr. Dermot Burns, 2017
- A comprehensive guide to writing a thesis.
Scientific Writing
- "The Science of Scientific Writing" by George D. Gopen and Judith A. Swan, 1990
- Quick read that gives tips to improve your scientific writing style.
- "How to write a good (enough) report by Andy Ruina
- Prof. Ruina's pragmatic take on writing with a goal of clear communication of your ideas.
Generative AI in Writing
Engineering Thinking
We ask you to write reports in school so that you learn how to write. You may not see it yet, but writing is also a key to improving your engineering thinking. The act of communicating each and every sentence, helps you think out the complexity of the engineering you are doing. If you let an LLM write for you, you remove a large amount of the thinking you need to be doing to become an engineer.
We cannot assess whether you can write if an LLM did any of the writing for you. We may be able to asses whether you can use an LLM, but that is not the same as assessing whether you can write.
LLMs can aid you during writing, in the same way discussing a paragraph with a person may aid you, but the words on the paper of your reports should be those that came from your brain and your fingers. When you are outside of school, you can use any tool you desire to produce the thing you are making. But in school, we are trying to get you to learn specific things and thus we put boundaries on what you can use in different tasks.
For example, you might be learning about integration of a mathematical function. Almost anyone can type a mathematical function into Wolfram Alpha and ask it to integrate the function but that does not teach you what integration is or means or what logic may produce Wolfram's result. You also have no idea if Wolfram gave you a correct answer or not. We teach you integration in almost the same way Newton and Leibniz taught themselves 400 years ago for a reason.
Generative AI is a tool that you can use to help you in your work. But it should not be used to directly produce content we use to assess specific learning objectives. The internship report, literature study report, and thesis report are tasks in your MSc designed to teach you to write through manual practice and feedback.
Writing in English
Using an LLM to help you write in English (whether it is a second language or your mother tongue) is useful. It can check spelling, grammar, translations, and other legibility aspects. We recommend using these tools to give you suggestions for such fixes and then you evaluate and incorporate them under your interpretation. Automatic translation saves times, but it does little to teach you a language, only using the language will. Once again, if the LLM does the writing for you, you lose the opportunity to learn how to write.
Take Grammarly, for example: it is a tool for spellcheck and grammar. It produces suggestions and explanations of in-line grammar fixes to help you to write correct and engaging texts while at the same time teaches you to write better. But it doesn't generate the whole text for you.
Lastly
If you cannot be bothered to write the text yourself, should your supervisors be bothered to read it for you? Your supervisors can just as easily talk to an LLM themselves if they want to read something it produced. Our role is to assess whether you can do technical writing and communicate your ideas and work yourself.
| Permitted | Non Permitted |
|---|---|
| type a paragraph, provide it to a LLM and ask for adjusments and feedback, manually rewrite the paragraph based on the LLM's feedback | type a paragraph, provide it to a LLM and ask it to improve the paragraph, paste the paragraph from the LLM into your report |
| use the LLM to translate a word, sentence, or sentence fragment as you are writing because you are struggling to get the write words and phrasing in English | write your report in your mother tongue, ask LLM to translate it to English, turn in the translation for assessment |
| ask the LLM to check the grammar in your report, manually fix grammar that is suggested by the LLM | pass your report to an LLM and tell it to fix all grammar and spelling, turn in the updated report |
Statement On AI Assistance
If you make use of writing tools that can essentially write text for you you must make a statement in your document explaining how you used such a tool. You should also be cautious when using such tools because the copyright of the text that it produces is ambiguous. The text is a derivative of other people's (copyrighted) writing and you may not want to laden your writing with possible copyright infringements. Nonetheless, make a statement on the use of the tool, for example:
During the preparation of this work the author(s) used ChatGPT 12x in order to improve the grammar in the writing. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication.
Common Issues
Pronouns
If you write in first person, you need to use "I" or "we". If you have a co-authored document use "we", i.e. "We measured twelve bicycles." If you have a single authored document (like your thesis), then it is generally not appropriate to use "we". You may use "we" if you are writing a tutorial-style document, where you consider the reader's to be part of "we", i.e. "Now we divide the metric by the mass of the person to normalize it:". This tutorial style may be appropriate in some places in a scientific paper, if you are considering "we" to be the author and the reader.
Conciseness
Conciseness typically equates to clarity. Edit things down, keep removing until only the essential information is present.
When to use a figure
If you refer to a figure in the text, that figure should be nearby the text so that the reader can read your interpretation of the figure while looking at it.
All figures should be referenced and discussed in the text, i.e. don't just throw in a figure because you made and it looks nice.
Only insert a figure when it enhances the story you are telling in the text.
Figure and Table Captions
Figures and tables and their associated captions should be able to be removed from the text and a reader should still be able to "read" the figure or table. This means that the caption should help the reader know how to read the figure. Anything that isn't obvious from the figure's design should be include, e.g. "The solid black lines represent the real part of the eigenvalue loci for the fifth eigenmode." The caption should not include discussion or interpretation of the figure, that belongs in your primary text.
Creating figures
Students often create figures that are saved from figure creation software without any adjustments to the default settings. The default settings are never appropriate. Search the internet for how to make publication quality figures with your preferred software. For example here are some
Python
- https://atchen.me/research/code/data-viz/2022/01/04/plotting-matplotlib-reference.html
- https://github.com/jbmouret/matplotlib_for_papers
- SciencePlots: matplotlib style to give well formatted plots for publication
Matlab
Common issues:
- The figure is poorly scaled for actual paper sizes.
- Bitmap images have either too high or too low resolution. Make your resolutions 300 dpi (dot per inch).
- Font sizes are too small on axis labels and legends, for example.
- Grey backgrounds are left in the figure (common for Matlab).
- Readers cannot discern individual lines or dots in the plot. For example, you may have a time series line plot of an accelerometer over 30 minutes. Showing all 30 minutes is simply a blob of color that is not useful. Show only 1 minute or 30 seconds so that the reader can see the data in a useful context.
Figure Copyright
In general, you cannot include figures you did not create draw yourself in your document. If you want to include the figure there are essentially three options:
- Ask the copyright holder for written permission to use their figure.
- If the copyright holder has licensed their figure with a Creative Commons license (or similar) then you can use the figure if you abide by the license's rules. This usually means displaying the reference to the license and the authors.
- It may be possible to use the figure without permission if the use falls under an exception in your jurisdiction. Dutch Law seems to have exceptions for "Illustration for teaching or scientific research", "Quotation for criticism or review", and "Use for the purpose of research or private study". https://en.wikipedia.org/wiki/Copyright_law_of_the_Netherlands
Math
Take time to typeset your math equations and number each equation for referencing in the text. Follow mathematical notation standards to make it easy for the readers.
Do not use \(a*b\) or \(a\times b\) to indicate scalar multiplication, \(ab\) is clear and more than sufficient.
LaTeX
This is how you do quotes in LaTeX!
`single quoted stuff'
``double quoted stuff''
Backticks on the left quotes are necessary to obtain correctly formatted quotation marks.
Prevent undesired linebreaks with the tilde ~. For example, you don't want a citation to be on a new line by itself at the end of a sentence.
This is my long sentence and this is a fact~\cite{JohnDoe2000}.
Code
There is no reason to include long scripts and programs in your thesis. If a product of your thesis project is code, it is best to archive your code in the proper file formats to something like Zenodo or Figshare and then cite in your thesis. You can also upload an archive of the code to the TU Delft thesis repository. It is appropriate to include code if you want the reader to read it. For example you may demonstrate an algorithm by showing a short snippet of code.
Appendices
Appendices are not just a clearing house to dump all the extra figures and tables you generated. Appendices are ancillary chapters and sections of your work. They provide supporting, but not necessary, information for the story in your main chapters. For example, if you say in a main chapter "We measured the stiffness of 10 bicycle tires and use those values in the model, see Appendix A." then appendix A should be a new section with written content that explains this measurement procedure and the results. If it was in the main text it would distract the reader from your main points, e.g. model description, but if the reader questions your stiffness values they can then read the appendix to see how you arrived at the values.
Referencing commercial equipment
You need to specify the product name of the equipment and any version specifications, then in parentheses list the company name, city, and country. Examples:
I applied a 6 Hz Butterworth low-pass filter to the signal using the butter() function available in Matlab 2019a (Mathworks, Natick, USA).
Each vehicle was equipped with a VR IMU BN0086 MEMs inertial measurement unit (Sparkfun, Niwot, USA).
File size
There is little reason for the PDF of a thesis to be larger than 10 MB. If your document is larger than 10 MB, then you have most likely embedded images that are much too large. Reduce the size of images before embedding and make sure to use .jpg images, rather than .png. A 300 dpi image that fits on an A4 page should only be a few hundred kilobytes.
Style
There are many writing styles; some styles fit with the norms in scientific writing and some do not. It is extremely important to make your academic reports and papers as easy to understand as possible. Some style choices will help you write more clearly.
Here are some recommendations:
- Write in active voice unless the context really needs passive voice to make things clearer.
- Write in present tense unless necessary to write in other tenses.
Allen Downey has some nice style notes: https://sites.google.com/site/allendowney/style-guide
Active Voice vs Passive Voice
You should write in active voice as your primary mode. Writing in active voice is generally simpler and clearer. Use passive voice sparingly. There should be a specific reason to use passive voice. Most major journals' style guides now request active voice. The Wikipedia article "English passive voice" is a good starting point to learn about the differences in active and passive and style recommendations. Allen Downey also has a couple quick reads that may help convince you of the merit of using active voice in scientific writing:
Initialisms and Acronyms
Don't use them. For every acronym or initialism you invent, it causes the reader to have to jump back to your definition every time they see it. The more you invent the more painful this is. There are two cases where it may be ok to use them: 1) the initialism or acronym is very commonly known to the expected audience, e.g. "PID" is an initialism that any control engineer would know, 2) you invent a single initialism or acronym for your paper due to repeating the phrase a very large number of times. Never use initialisms or acronyms in titles or abstracts. Always define any initialism or acronym (that your or others invented) on the first use of the phrase. If your sentences have more than one or two acronyms or initialisms present, you should likely write the phrases out to ease reading. Initialisms and acronyms make it easier for the writer but not the reader.