Artificial Intelligence #44: How to choose a good research topic ..

Welcome to Artificial Intelligence #44

Since I span industry and academia, I see this question often - How should I choose a good research topic?

I wish I had understood this much earlier in my career! Its useful even if you are not strictly a researcher. I think its very important, especially today, to come back to a more scientific / rational way of thinking

Early in my career, I heard a well known academic say: The world wide web is not a good research topic. I was intrigued by this. So, his response was: yes it is a game changer - and it is practical - and it obviously works - but it is not a good research topic because it does not have an underlying theoretical foundation.

Yes, of course we can find a number of theories covering aspects of the WWW but this response needs some exploring.

So, recently when another student asked me for advice on choosing a research topic - I provided the following explanation.

I hope it will help you also

The explanation is based on two excellent resources

First, Selecting a Research Topic: A Framework for Doctoral Students - Andy Luse, Brian Mennecke, Anthony Townsend

This, I highly recommend, whether you are as researcher or not

I summarise some key ideas from this paper as follow

• When setting out to develop a research project, many doctoral students begin by jumping straight to a solution to a problem before they have given the problem good thought or sometimes before they even have a well-defined problem at all.
• The risk that this strategy creates is that the researcher will not fully understand the problem and, in turn, might fail to recognize fundamental issues that frame the context of the problem.
• To prepare oneself for identifying a research topic, a student should be willing to challenge previously held beliefs. While this can be difficult, there are some techniques that a student can use to foster creative thinking. This takes some time

Kuhn (1962) delineates three methods for approaching a problem that allows the researcher to understand the problem more fully before attempting to develop and implement solutions:

Isolate and Give Structure: The researcher should isolate the problem from other external factors to gain a greater understanding of the problem itself, specify and define concepts within the problem, clarify levels of reference, like part-whole or micro-macro concepts, before specifying relationships between key concepts and organizing and categorizing these concepts into an overarching typology.

Magnify the Problem: Magnification focuses on a particular isolated section or sections and amplifies every portion of this section to allow for a more thorough understanding of that specific isolated piece of the problem. This may involve reading about the problem portion, performing experiments to reproduce the problem portion, or merely thinking more thoroughly about the problem portion (e.g., thought experiments).

Search for Theory: Apply one or more relevant theories by conducting a complete literature review. Several guidelines should be considered when conducting a literature review. The primary goal of a thorough literature review is to find sufficient relevant theory and research to formulate a well-structured argument from which your research questions can stem.

Theory development

One choice to be made when developing a theoretical base for a problem is the theory development methodology to utilize. A theoretical basis can be intricate in detail and pertain to an already established area, or it can be completely novel and address new or emerging domains.

A second element important to researchers and particularly relevant for beginning researchers is the concept of contributing knowledge. Contribution to knowledge implies an increase in our understanding of a phenomenon to contribute to the overall body of knowledge in the area.

Thus, the key is, for both creating a theoretical basis for the problem and for contributing to the 'state of the art'. Hence, the need for a theoretical foundation for a research problem

The best way to think of this idea further is to explore the ideas of Thomas Kuhn referred above in his seminal book on Structure of Scientific Revolutions

A good analysis of Kuhn's thinking is from Thomas Kuhn - Science as a Paradigm

In essence, Kuhn helps you to understand science and how science evolves. That is why you need a theory/ foundation for your thinking.

According to Kuhn, a theory is

1. Accurate – empirically adequate with experimentation and observation
2. Consistent – internally consistent, but also externally consistent with other theories
3. Broad Scope – a theory's consequences should extend beyond that which it was initially designed to explain
4. Simple – the simplest explanation, principally similar to Occam's razor
5. Fruitful – a theory should disclose new phenomena or new relationships among phenomena

Kuhn's key idea is

• Thomas Kuhn argued that science does not evolve gradually towards truth.
• Science has a paradigm which remains constant before going through a paradigm shift when current theories can’t explain some phenomenon, and someone proposes a new theory.
• A scientific revolution occurs when: (i) the new paradigm better explains the observations, and offers a model that is closer to the objective, external reality; and (ii) the new paradigm is incommensurate with the old.
• Scientists have a worldview or "paradigm". A paradigm is a universally recognizable scientific achievement that, for a time, provides model problems and solutions to a community of practitioners.
• A paradigm is a basic framework of assumptions, principals and methods from which the members of the community work. It is a set of norms which tell a scientists how to think and behave and although in science there are rival schools of thought there is still a single paradigm that all scientists accept uncritically.
• Scientists accept the dominant paradigm until anomalies are thrown up. Scientists then begin to question the basis of the paradigm itself, new theories emerge which challenge the dominant paradigm and eventually one of these new theories becomes accepted as the new paradigm.

According to Kuhn, Science evolves in phases

Phase 1: Pre-science: The pre-paradigmatic state refers to a period before a scientific consensus has been reached.

Phase 2: Normal Science: (most common – science is usually stable) - where a paradigm is established which lays the foundations for legitimate work within the discipline.

Phase 3: Crisis: when a paradigm shift occurs. An entirely new theory/idea emerges - much like Einstein challenging the Newtonian physics prevalent in the day.

Phase 4: Revolution: Eventually a new paradigm will be established,

So, I hope that gives you an idea about how to think of research topics and why a lot of what we see, cannot be a good research topic - because the theory is important in scientific foundation.

Like i said, I wish I had understood this much earlier in my career! Its useful even if you are not strictly a researcher. I think its very important, especially today, to come back to a more scientific / rational way of thinking

I hope you find this useful also

With contributions from Diana Teresa Parra Sanchez

Image source Structure of Scientific Revolutions -