Is correctness necessary?

  • If the goal is acquiring knowledge, then it is necessary.

  • If the goal is inspiration, then it’s not necessary.

  • Direction

  • Thinking while writing Process

    • Influencing thoughts through the editor
    • Who do you want to read it?
    • Self-dialogue
  • Real-time conversion of meeting minutes

  • Transforming it further when a different perspective is already provided

  • Assuming a situation of either alone or with a mentor (brainstorming)

  • Using humans instead of LLM for filling in the gaps?

This week

  • Approaching abstraction
    • Methods other than using LLM
    • Survey
  • Proposing/automatically selecting what kind of conversion to make (source/target)
  • Using this to write down what you usually think about

Is there a role like a translator of concepts?

  • Good at analogies, lateral thinking
  • Transformation of historical-modern like Koten Radio

  • Approach to abstraction _
I see, you want to abstract while maintaining the sentence structure and avoiding excessive information omission. In this case, the following approaches may be effective:

Part-of-speech-based abstraction:
"Apples are delicious"
→ "[Noun] are [Adjective]"
Semantic category-based abstraction:
"Apples are delicious"
→ "[Food] are [Sensory evaluation]"
Grammar function-based abstraction:
"Apples are delicious"
→ "[Subject] are [Predicate]"
Semantic role-based abstraction:
"Apples are delicious"
→ "[Object] are [Attribute]"
Combination approach:
"Apples are delicious"
→ "[Noun: Food] are [Adjective: Sensory evaluation]"

By using these approaches, you can achieve a certain level of abstraction while maintaining the sentence structure. Additionally, by adjusting the level of abstraction, you can control the amount of information retained. When instructing such abstraction for LLM, you may use the following notation:
Copy<abstract_sentence>
[Word1:{Part of speech}:{Semantic category}:{Grammar function}:{Semantic role}] 
[Word2:{Part of speech}:{Semantic category}:{Grammar function}:{Semantic role}]
...
</abstract_sentence>