Paper Clinic

In these sessions, senior researchers like Prof. Albrecht Schmidt introduced the basics of writing good research papers through interactive discussions. They focused on: 1) how to write a good research paper, and 2) how to do systematic literature reviews. Afterwards, PhD students presented rejected/to be submitted/planned paper (ideas) to us. Fellow participants helped in finding the caveats and ways to improve the paper.

How to write a good research paper?

  1. Write down idea on the first page before doing the actual research -> shape research idea and study
  2. If ethics is required, writing it up is pretty equal to structuring the paper
  3. If you have a prestudy/main study running/incremental study write it up while it is running -> Methodology documentation
  4. Be clear and focused within the content of the paper
  5. Emphasize what is unknown and what is your contribution
  6. Writing up papers is a primary mechanism to do and conduct research apart from reporting the results
Design your papers with an exciting idea which goes through the paper
  1. Be passionate about your research idea and question
  2. Writing a paper is how ideas are developed in the first place
  3. One contribution, idea per paper (“sharp and clear”)
  4. Be explicit and try to explain the idea of your paper. If this does not work, it is hard to convey the idea to readers
  5. The reader should have no doubt about your research idea
Tell a story
  1. Explain the idea
  2. What is the problem?
  3. Explain why the problem is important
  4. How do you solve it?
  5. Walk through your idea and problem-solving strategy
  6. Pinpoint the problem
  7. Show how your idea works
  8. Distinct your solution to existing alternatives if there are any
  9. Convince people that your solution is working/valid (Method, study, evaluation) State your contribution!
  10. Show how your study supports the claim your paper makes -> showing the validity
  11. Reviewers should understand your statement
  12. When preparing the paper: Write the abstract first before doing the study -> Think about if/why your claims are substantial
  13. Most people will read the Title and Abstract (in HCI)
  14. Show in related work how other people proved their claims
Nailing your contributions to the must
  1. Describe the problem and state your contributions. Try to keep the introduction within one page
  2. Highlight the problem with examples and incorporate in the introduction
  3. “You have a problem, and I have something to fix that” – Albrecht Schmidt
  4. Be concise
  5. State your contribution
  6. The list of contributions drives the entire papers -> Write the contributions down with bullet points
  7. The contributions should be specific and concise
  8. The introduction makes
  9. The rest of the paper provides evidence for your claims
  10. Frame the claims -> Show how your results can be applied to a general audience/use cases
Related Work: Later
  1. Related work should not be boring the reader
  2. Suggestion: Put related work later in the paper, however, it is safer to go along with the community (RW after the introduction)
  3. Highlight the delta between your work and previous research
  4. Do not make look other’s work bad!
  5. Credit the people who have done similar work
  6. Understand what you put in the related work section -> Discuss it and show how your work is different!
  7. People who are referenced are very likely to be the editor/AC/Reviewer
  8. Don’t only cite Google Scholar page 1 when researching
  9. Explain the content of related research briefly
Put your reader first and listen to them!
  1. First reader are the reviewers
  2. Let experts read your papers first (colleagues or peers)
  3. Keep the length restriction and paper format!
  4. Provide visual structure to the paper -> Be nice to the people who read that
  5. Use simple direct language. Be concise!
  6. Writing paper requires practice and experience from postdocs and peers
  7. Collaborate with people in your field
  8. Keep the abstract to one paragraph -> describe what was done and what was found!
  9. Abstract structure:
    1. What is the problem?
    2. Why is the problem important?
    3. What is new about what you done?
    4. What did you find?
    5. What does it implicate in the bigger picture?
Further reading
  • How to write a good research paper by Simon Jones, Microsoft research (slides)

How to do systematic literature review?

In this part, Prof. Schmidt used one of the presented student talks about “applications of virtual reality in treating anxiety disorders”  as a probe to show how to do a systematic literature review.

  • If you look at google scholar for “anxiety disorder virtual reality”,  you get a lot of results, from various research domains.
  • Therefore, we take a shortcut by searching the ACM digital library. Limit your search to computer science.
  • Afterwards, we repeat the search process  in the host-libraries for the publications (e.g. Springer, IEEE, Microsoft Academic)
  • Next, we look at google scholar.
How to search a host library?
  1. Use full-text search
  2. Reduce search by adding + to each term
  3. You list everything you find
  4. If it is a low number (e.g.8) -> Read the abstracts -> Write in own words what they did
  5. After gathering a large number, re-read and start categorizing
  6. Go systematically and document what you are looking for
  7. The number of citations makes it more relevant
  8. Reduce the years back, to get the most relevant niche applications nowadays

EXAMPLE

  • ACM : search for (+virtual +reality +therapy) “notice the + operator”
  • Refine by publications domain
  • Dig down deeper: look for (+anxiety +disorder)
  • IMPORTANT! make a table -> get 30 most important -> classify them
  • Find a fundamental publication (very cited), go through all the papers that cite it

Your objective is to find a combination of keywords where there are publications but not too many. If it is in the order of thousands, it is dangerous to research there as it is an exhausted area. It is better to find a niche than not finding it, even if this niche is narrow

Overview on discussed papers

Paper 1: By Matthias Hoppe
  • Feedback in VR comes from controllers
  • How can we mimic objects?
Paper 2: By Alexandra Voit
  • Calendars are used to organize the daily lives
  • Development of an ambient and pervasive digital wall calendar with event suggestions
  • Evaluation of the developed system in a four-week in-situ study (deployed in the users home)
Paper 3: By Sebastian Feger
  • Visualize contributions from research preservation
  • How is the contribution of physicist viable?
Paper 4: By Nada Terzihemic
  • 14 participants journaling what they ate with context and pictures
  • Captured windows of opportunities for suggesting food intake
  • How could the existing dataset be used to construct a conference/journal paper
Paper 5: By Pascal Knierim
  • How can AR be used in physics?
  • How can it be used in education?