Scientific Reading, Writing and Presenting

9 minute read

Published:

This is a note on how to do science reading, writing, and presentation.

Scientific Reading, Writing and Presenting 1

How do I find relevant literature?

If you don’t have a starting point (paper) yet, use keyword based search

If you found an interesting paper, you can use forward and backward search

  • For finding new papers that cite this paper, click the “Cited by” on Google Scholar
  • For finding older papers cited by a paper, read related work section (references)

Use online tools

Is a paper worth reading?

  • In general, it’s difficult to say before reading
  • You must invest time, it’s a hard work

How do I organize my literature?

Citation and reference management software

  • JabRef (offline)
  • Mendeley (offline / online)
  • Zotero (offline / online)

Scientific Reading

How to read a paper?

  • Unless 100% sure the paper is relevant, don’t read it linearly from start to end
  • Instead, take a quick look at abstract, contributions, teaser, results (~ 10 min)
  • Take notes, summarize and decide if you want to read it in depth (~ 2h)

Keep notes

  • Highlight important passages
  • Add comments, take notes
  • Mark what you did understand and what you didn’t
  • Summarize in your own words

Understand the content deeply

  • Look up unknown concepts
  • Read prior work if necessary

Scientific Writing and Reviewing

How to write a report?

Start early and iterate

  • Writing needs time
  • Ideas form while writing
  • Problems surface while writing
  • It’s important to start writing early on and iterate
  • Start with a longer text and shorten it later
  • Be inspired by the papers you read, adopt good writing styles
  • Make sure language and grammar are correct, avoid colloquial language (use tools like Grammarly for finding and fixing typos)

Come up with a good structure

  • Abstract - Motivation and key messages? (200-400 words)
  • Introduction - What is the problem? Where does it occur?
  • Motivation - Why should we care? What applications?
  • Background - Which background knowledge is necessary?
  • Approach - How does it work? How do different methods compare?
  • Results – What has been achieved? What works and what doesn’t? Why?
  • Summary – What should we have learned? Future work?
  • References – Use the \cite command

Follow good scientific practice

  • Your text should be your own exposition and explain things in your words
  • Don’t copy sentences 1:1 from your sources (unusual in natural science)
  • Whenever stating a fact that is known, add the corresponding citations
  • Make sure all related work is cited appropriately (otherwise: plagiarism)
  • Citations are added before punctuation marks
  • Use Latex in combination with Bibtex to manage your citations and bibliography
  • Use the cite package to format the bibliography alphabetically

Figures help understanding

  • Place figure outside running text, usually at top of page
  • Adjust font size to font size of main text
  • Caption should describe figure concisely to be understood stand-alone
  • When using a figure or table from another source, cite the source in the caption
  • Make sure all figures and tables are referenced from the main text
  • You can reference the same figure or table multiple times

What is a review?

Reviews are a core element of the scientific process

  • Provide feedback to the authors prior to publication, including: Language, clarity, math, rigor, references, experiments (and novelty)
  • They serve to improve the manuscript ⇒ make concrete suggestions
  • Reviews are objective and state both pros and cons
  • Reviews judge the quality of a paper and if it gets accepted
  • Area chairs / associate editors make final decision based on reviews
  • Often the authors and reviewers don’t know each other (double blind)
  • Sometimes the reviewers can see the author’s names (single blind)

Your review provides value

  • What is the report about and why is the problem relevant?
  • Is the structure of the report appropriate or could it be improved?
  • Is the problem description clear? Is it easy to follow?
  • Are related works cited appropriately? Are citations missing?
  • Are the related works clustered in groups / a taxonomy? Can it be improved?
  • Does the report discuss pros and cons of the presented methods?
  • Is the problem described formally? Is the math correct? Can it be improved?
  • Are all symbols / formulas explained? Common notation? Easy to follow?
  • Are the methods well illustrated? If not, what is missing / what can be improved?
  • Do the table / figure captions help understanding the tables / figures?
  • Is the the language clear and correct? Are there typos?
  • Is the experimental setup properly explained?
  • Are all utilized datasets, baselines and metrics explained?
  • Does the report present qualitative and quantitative results?
  • Are the results discussed in an objective manner (pros and cons)?
  • Provide both positive and negative feedback, be constructive

Where should I start?

  • LaTex is a standard tool for academic typesetting
  • Latex allows for professional typesetting of text, formulas, illustration and tables
  • Overleaf is an online Latex editor: no installation required, good for beginners

Scientific Presenting

What should I pay attention to?

  • Remember that you know more about your subject than anyone else
    • Never assume knowledge of your audience
    • Pitch your talk such that experts and non-experts (your friend) can understand
  • Come up with a good structure
    • Introduction – What is the problem? Why should we care? Applications?
    • Approach – How does it work? How do different methods compare
    • Results – What has been achieved? What works and what doesn’t? Why?
    • Summary – What should we have learned? Pros and Cons? Future Work?
  • Finalize your talk well in advance

How should I prepare my slides?

  • Your story matters
    • A common mistake is to create too many slides
    • Make sure you spend at least 1 minute per slide (less slides than minutes)
    • Keep slides simple and clean to not distract the audience from your message
    • Use a simple template, avoid excessive decoration / logos
    • Use a large font size
    • Use simple and consistent color scheme
    • Use a homogeneous background
    • Add your name, presentation title and page number to the footline

Focus on the Core Message

  • Simplicity is key
    • Remember that you can convey only 1 message per slide
    • Use maximally 3 bullets per slide, short sentences and simple figures
    • Avoid big tables, try to use simple diagrams or bar / pie / curve plots instead
    • Highlight important parts of a diagram / illustration as you go through it
    • Avoid long text, use keywords instead and explain while presenting
    • Your slides should support and illustrate your message, not replace it
    • But it is important to explain all baselines / metrics / symbols that you show
    • Avoid animations unless important to ease understanding (but then use them)

Language

  • Your language matters
    • Make sure language and grammar are correct
    • Use simple yet precise language
    • Avoid colloquial language
    • Avoid unnecessary complicated language
    • Use tools like Grammarly for finding and fixing typos

Formulas

  • Formulas are tough
    • Formulas are challenging as they can become complex
    • However, they can help to make your message more precise
    • Distill the core ideas into the most relevant formulas
    • Remove unnecessary symbols (indices, etc.)
    • Make sure your notation is consistent across your presentation
    • Explain mathematical content slowly and step-by-step
    • Show formula side-by-side with a simple example
    • Often its easier to understand by example

Which presentation tool should I use?

  • Use any tool you get along with (PPT, Beamer, etc.)

How should I practice my talk?

  • Practice is everything
    • Present your talk to yourself several times before giving the talk
    • While practicing, keep track of your time and adjust presentation
    • If possible, present to others with no expert knowledge to get feedback
    • Avoid using presenter’s notes (and under no circumstances read them)
    • Plan your presentation without laser pointer or file cards
    • Instead, practice free speech and highlight items directly on the slide
    • Show bullets / illustrations one-by-one, highlight key numbers / plots, etc.

What should I do if I get nervous?

  • Being nervous is normal
    • Adrenaline helps to be present in the moment
    • Setup your laptop prior to your session and make sure video / audio works
    • Try to remember you first 3 sentences by heart, this helps getting started
    • Avoid using presenter’s notes (and under no circumstances read them)
    • Instead, practice free speech, every time you will become better

What should I do during the talk?

  • Capture your audience
    • Keep eye contact with your audience
    • Speak slowly, pause from time to time (i.e., after important points)
    • Explain the content of every slide, do not leave things out
    • Avoid using pointing devices (mouse / laser pointer / hands) too much
    • Be excited about the topic that you are presenting (fosters interest)
    • Keep track of the time (put a timer, e.g., mobile phone, in front of you)
  1. The original author is Prof. Dr.-Ing. Andreas Geiger, University of Tübingen / MPI-IS.