Hints for PhD Students

H.Mössenböck

Writing a PhD thesis is research. You have to demonstrate that you are able to make original contributions in a non-trivial area and to push the state of the art in this area a bit further. This involves that the techniques suggested in the thesis are sufficiently novel. Only very few dissertations are fundamentally novel. If you achieve that: congratulations! However, the thesis should at least contain a few novel aspects which make it worth a dissertation.

Here are a few hints for planing and performing your research:

Rule #1

  • Read this document once a month ;).

General hints

  • Look for real problems in practice and try to solve them. Reading papers is important, but you should not only do another variation of a solution from literature.
  • Look for people who work on similar projects (look at the internet or at conferences). Try to exchange ideas and to discuss your work with them.
  • Carefully study the state of the art and then write up a research proposal. What are the research questions or the problems that are not sufficiently solved so far? How would you approach these problems? What would be your original research contributions? How would you evaluate your approach to show that it works correctly and efficiently?
  • Before you implement anything, try to apply your ideas manually to some sample problems of different size.

Literature Survey

  • Read as many papers as possible in the area of your research. At the end of your PhD studies this list should at least contain about 100 papers. Exchange this list with fellow researchers.
  • Summarize every read paper in your own words and add these summaries to your literature list. Point out and collect ideas that might be useful for your own work.
  • Regularly search the web for keywords in your area and see if you find new interesting stuff.
  • Make a list of all relevant journals and conferences in your area. Regularly search these media for new papers that are relevant for you. Use the references at the end of every read paper to find more relevant papers.

Notes

  • Keep a list of questions that you want to look at. What are the problems to be solved? Update this list regularly and check if you make progress, i.e., if you have solved some of the problems (possibly triggering new questions and problems).
  • Keep a list of ideas. Write down every idea, even if you don't have time to try it out right now. You get ideas from reading papers, from attending conferences, or from discussions with colleagues.
  • Keep a project diary with daily entries shortly describing what you did on this day.
  • Keep a to-do list. Which aspects of the implementation are still missing? Which questions would you like to look at more closely? Which papers do you still have to read?

Publications

  • You should write approximately one paper per year summarizing the work that you did and highlighting its contribution to the state of the art. This will lead to about 4 papers during your PhD studies. Usually, the thesis is then just an integrated version of all the papers that you published.
  • When writing something up, it is often easier to start with a tech report. A tech report is not restricted by its size (as many conference papers are). It need not be as refined as a real paper, but it helps you to collect your ideas. A tech report can easily be turned into a paper, and the quality improves when it is rewritten for publication.
  • Keep a list of all relevent conferences with their submission deadlines. You should explicitly target a specific conference and start with the paper early enough (at least 2 months before the deadline).