1. Define what the problem is -> Create the main topic
2. Break down the thinking process for solving the problem into small pieces and solve them as modules -> Break it down into subtopics
3. As you search for solutions and pose questions, new small subproblems or definitions of problems at the same level emerge
4. Repeat steps 1–3

Motivation

While extracting insights from text, I realized that extracting insights has to begin with asking good questions. When we keep extracting this and that because they seem like they might be necessary, without even knowing what the problem is, we often lose the direction of the project and find ourselves no longer knowing what we are doing. As we proceeded with the project like a racehorse looking only straight ahead, we ended up not even knowing what we were actually doing.

I concluded that this concern stemmed from ignorance and looked for a solution. I explored how text mining is done and, more broadly, how people who perform data mining think.

Conclusion

Let’s devote more effort to identifying what the problem is

The conclusion I reached was that we should devote more than 90% of our time and effort to identifying what the problem is, and use 10% to find a solution. This does not apply only to analysis and extracting insights; it also applies to coding, learning new concepts, and the process of getting to know myself.

In the act of finding a solution, everything begins with defining what the problem is. Good questions also begin with defining what the problem is.

Then how do we determine what the problem is? We also need to think about how to define what a good question is. This is a new problem for me, because “too ambiguous” and the conditions for a good question involve various factors. My conclusion is that, ultimately, I should spend a great deal of time defining the problem, look for solutions, and try them; if the problem is solved or remains unsolved, I should then devote myself to defining a new problem based on data from the previous attempts. Since this problem definition becomes a clue to what I do not know, it seems to be a very important habit for learning as well.

Side note

I came to think that in order to learn from people who are good at something, I need to spend time defining the problem, and if I do not understand the problem, I need to make my questions more specific and sophisticated. There were also people who thought they had become good at something for reasons other than the core reason, even though there was a real core reason they had become good at it. Of course, they became even better because the peripheral reasons they improved were combined in collaboration with the core reason, but when benchmarking someone who has become good at something, the first step is to find the core reason that made them good, and then to discover the peripheral reasons. It is also important to apply this to myself.