Static model structure / archiving

[sudojunior]

An idea to completely decentralize the data output.

Current ideas have a *.markov file that would be both easy to generate and parse.

Comments can either be added through an interface in the library, or manually - as long as it doesn't directly affect how content interpretation occurs for a specific model.

Based on partial input from The Zygon Inversion (Doctor Who):

# [Notes]
# This is a model selection of the actual speech, the end goal is to run this speech through using a dialogue parser.
# Currently, all formatting (difference between upper and lower case, any use of punctuation or phrase emphasis) is removed for the purpose of outcome consistancy.

[The Doctor]
this: a, funny, is, --, no, pain
a: war, bigger
war: this, i, than
funny: little
little: thing
thing: this
is: not
not: a, on
fought: in
in: a
bigger: war
than: you, anyone
will: ever
ever: know, imagine, be, have
know: i, what

[Bonnie]

Consider using a literal space instead of commas to allow full formatting to be implemented. (#5)

// This is a literal representation (in graphql) for how a line of dialogue could be seen by an interpreter.
class DialogueLine {
  string speaker; // AKA author
  string line; // AKA message
}

Other references

• Transcripts Archive
• Dialogue Structure from ~/BuildingBridges at master
• TV & Movie Transcripts - Forum Index
• SimplyScripts

[sudojunior]

Additional stat tracking opportunity to track individual word relationships by providing numbers to each word and how many times they were linked, rather than vice versa to count how many times a word was called upon.