Guidelines for a Research Plan

The goal of the Artificial Intelligence Laboratory is to foster intelligence in all its facets by promoting excellence in basic research, education, and society at large. With our activities we hope to contribute – in small ways – to making the world a better place in the 21st century.

As such we expect high-quality research from every PhD student in our group. Your work needs to be technically grounded, scientifically solid and significantly evaluated. We want you to become an internationally known expert in your particular area of research. As an AI Lab researcher, that needs to be your goal!

• Start with a state-of-the-art report in your field of interest
• Distill open research problems in the field
• Formulate the research problem that you want to address
• Layout challenges, pitfalls, and potentials of solving this research problem
• Propose a possible solution and how to get there
• Decompose your plan into smaller pieces, i.e. work packages and milestones
• Develop a time plan and a publication (dissemination) plan
• Start with workshops, then conferences, then journals
• Carefully (and early on) address how to evaluate or validate your work
• Always have an executive summary of your research available
• Evolve the research plan - it is your working document for the whole PhD study

• Have your research plan within one year upon start of your PhD study (+/- 6 months)
• Use proof-of-concept prototypes early on. They help you develop the concrete research ideas
• In the first year, international workshops are good targets for early ideas (that work)
• Get early and regular feedback from your advisor and senior researchers in our group
• Ask for one-on-one meetings - your advisor and senior researchers will be happy to give you feedback
• Get involved also in co-authoring papers together with senior researchers - it helps developing paper writing skills!
• If you don't have an accepted research proposal or a single paper about your research work published by the end of year 2, that's a sign of alert!
• Your PhD research should not take longer than 4 years to get it done!

The research plan needs to have a title page including your and your advisor's name. The research plan needs to be signed by both you and your advisor at the end of the document.

The research plan should contain a short summary to identify the area of your research, the main goals and the expected contributions to your field of research.

A research plan needs to have a vision: what particular problem are you trying to solve? what will be the impact on the discipline? what is the delta or the added value of your research work? These questions need to be answered in the plan.

A brief introduction to the field of your research and the issues to be addressed should be included. It is important to already make clear what the motivation, the general methods and outcomes of your research will be. For example, will you be designing an algorithm and proving it correct or developing a framework and designing experiments to evaluate it?

The goal of this section is to describe the setting of your research in terms of previous and related work. It should explain the significance of your research and identify research challenges. In a summary document such as this, you cannot go into details of all related work, so it is important to focus on the main concepts and to reference key papers. It should convince the reader that your research is worth doing and also that you have the necessary background knowledge to carry out that research.

This section should include the main goals of the thesis and then provide details of the following:

• the research questions to be addressed
• the methods to be used
• the expected contributions

Your work plan should be presented in terms of work packages. Each work package should represent a major piece of work. The number and nature of the work packages will depend on the topic of research and this is where the student should be guided by their supervisor. As a general guideline, a work package should describe a min of 3 and max of 12 months of work. Each work package description should define the task and the expected outcomes.

With reference to the work plan, you should describe the progress made to date with details of any results, including publications accepted or submitted.

Your research work needs to be evaluated and you have to plan that ahead of time. You need to consider which kind of evaluation or validation to run for showing that you achieved your goals, be it an empirical study, a case-study based validation, a controlled user study, etc. In this section you should layout which experiments you plan, how they will look like and what you will validate with them.

A time schedule showing clearly when you started your PhD, the work packages completed or under way and the envisaged start and end times of the future work packages should be included. This should include estimates of when the thesis will be completed, including the writing of the thesis report.

A list of papers referenced in the description of the state-of-the-art should be included. Since the research plan is intended to be a summary document, it is not necessary to include large lists of all papers that you have read, but rather to reference key papers in your area of research.