Handeling language in computers

Information from language

  1. Tokenizing
    Dividing into group of letters (a-z) and special characters (!,-)
  2. Classifying tokens
    Loop until no new discovery:
  3. Setting up entities with modifiers
    Words should be stored in their most primitive form. Any change from this should be stored with referred modifiers. This way assosication is kept.
  4. Applying attributes
    Adjectives and adverbs should be related to the correspondent word as attributes.
  5. Setting up relations of subject and object
    The relations should be distinguised from attributes.
  6. Context references
    One for each sentence, sub-sentence, paragraph, section etc.
  7. Resolving 'it', 'this', 'that' etc.
    If not emphasised or given context - previous context. Higher emphasis - higher up in the context hierarchy.
  8. Apply references for abstractions and realizations.
  9. Apply references for containers and parts.
  10. Apply references for intention and solution.
  11. Apply references for causal and effect.
  12. Apply references to time.

Coherence from information

This can be run independent from the rest by only adding new references and not modifying anything.

Queries of coherence

  1. Finding the nodes in the question
    1. Map the query to the existing data.
    2. The answer to the question is the mapping to the question word (e.g. 'what', 'why').
  2. Counting and equality
    Where several solutions partially apply, the one that matches the most shall

Learning grammar

By rules or by patterns. The last might be somewhat adaptive by asking when there is a new pattern.

Implementation


Valid HTML 4.0! Last updated 2003-01-23.