What is Scientific Research? What should I do to start my research?

A large number of researchers including myself start their research without enough preparation. They forgot to think about what is research and how to conduct a good research before they set out to do it. This will plant a big problem for future research. This article will analyze the composition of scientific research and provide some advice before starting research.

Composition of scientific research

Scientific research is a process of exploring theories, phenomenon, or hypotheses. In the area of computer science, it usually consists of five stages: targeting problems, proposing solutions, implementing prototypes, doing experiments, and analyzing results. According to the stages, the first stage is likely to be ignored most in practice. Researchers often start with proposing solutions even directly implementing a prototype before targeting problems. As a result, the research always has unclear motivation. The motivation explains the reason for the research and shows the contribution of the work. An unclear motivation even influences the value of the work. Therefore, the lack of motivation is one of the most comment reasons why some awesome papers are rejected by the reviewers. To conduct high-quality research, we need to properly target the existing problems.

How to target problems?

First of all, you need to have a good understanding of the area. The word “good” means “deep and narrow” rather than “shallow and wide”. You need to review a large amount of related work in the specific research area. Some old and widely cited paper is a good start point. After reading these classic papers, you should turn your head to the latest publication on the well-known conferences/journals. These papers will help you build a state-of-the-art structure of the area. The next thing is to find out as many research gaps as you can. The research gaps are some small problems that you think can be improved. To be noticed, these gaps are not your final research question and you are not supposed to fill all the gaps. The final question will be selected from the research gaps. It can consist of one or multiple research gaps that you think need to be filled urgently. The form of the research question should be similar to a hypothesis. The subjects, objects, and conditions of the research need to be described specifically and clearly. The following examples are well-defined and vague research questions respectively.

How does head position context awareness reduce the GPU consumption of an autostereoscopic display system?

A well defined research question

How to improve an autostereoscopic display using context awareness?

A vague research question

As we can see from these examples, to propose a good research question, unclear concepts and statements should be avoided. After making sure the question is clear, you can start your research!