Tips for Research

Tips for Research:

1.    Research vs. development: Research is the non-trivial issues found during the development process which might be planning, building, or troubleshooting a system.

2.    Problem matters most: A research topic has a generic problem statement with (1) inputs, (2) outputs, (3) objective function, (4) constraints, and (5) assumptions, where inputs are the given info that you can collect, outputs are things you want to decide, but subject to the constraints, so that the objective function can be maximized or minimized. The decisions are made under certain assumptions.

3.    Survey before solution: A solution to the defined problem statement is a method, mechanism, or algorithm that can optimize the objective function. There could be some existing solutions already. Thus, before developing a solution of one's own, one needs to do a survey to find and compare existing research works and argue why their solutions are not good enough and the society needs a better solution. Usually a survey would have a comparison table identifying the ideas, pros, and cons of existing methods and yours.

4.     Evaluate your solution: A research is incomplete if without evaluation of the proposed solution to the defined problem. The evaluation would have experiment setup and experiment results, where the former is to describe how the experiment is conducted with certain parameter configuration such that readers of this work can reproduce the results. The experiment results, usually presented in curves or bars in figures, should answer how effective the proposed solution is. The results should be better interpreted with three kinds of observations: (1) how the curves/bars look like (grab the key numbers that yourself and readers could remember), (2) why they look like these, (3) insightful implications to us or stakeholders.

5.     Generic template of a thesis/paper: A research is documented in a thesis or paper with the following components: (1) title (should be clear and sexy), (2) abstract (a summary of the entire work, with issues, previous works, our proposed idea why it could work better, results with key numbers), (3) introduction (with motivation, issues, brief survey, our approach, organization of the thesis or paper), (4) background (tutorials, surveys, i.e., related work), (5) problem statement, (6) solution approach (with a sexy name of your method/solution), (7) evaluation, (8) conclusions and future work, (9) reference.

6.     New problem vs. new solution: Throughout the research period, defining a problem statement and survey existing works are critical because they determine the impact or value of your research. A problem well defined is a problem half solved and with its impact almost determined. After that, developing a solution and evaluating the solution are the other critical parts. Usually a new problem with new or even old solutions has more contribution than an old problem with new solutions.