conceptology

These are evolving ideas to improve Wikischool with a better conceptual and semantic interface.

The Explore app (info) is the first development in this direction.

• Better semantic user-interaction with the text data, eg.: “understand natural language user queries better and formulate synthesized answers.” Create a simple user-language for creating concept queries.

• Interactive, audio-visual storytelling based on concepts, internal/external content, a good TTS system and voice input.
  • Can we try to achieve a Human toolset for wholesome infotainment on many subjects?

• Better interpretation, logic and visualization of the concept lists and the individual concepts.
  • concept inception –> forms –> concept extraction
  • idea-class versus form-instances

• Semantic sentence embedding encoding-decoding:
  • demo: semantic book search
    • universal sentence encoder
  • video: “Representations for Language: From Word Embeddings to Sentence Meanings”
  • video: “BERT NLP intro
  • NLP research overview

• NLP meaning visualizations:
  • embeddings (“mapping from discrete objects, such as words, to vectors of real numbers”):
    • video: visualizing high-dimensional spaces
    • tensorflow embeddings projector (code)

• Concept modeling experiments:
  • (to research)
  • ”A Conceptual Space Logic“ (pdf)
  • Metamath.org (tiny language that can express theorems in abstract mathematics)

• related keywords: conceptology, concept logic, concept heuristic, ontology, semiotics, Curry–Howard correspondence, semantic web, automated text classification