I had a conversation with a friend of mine many years ago about how it could be possible to simulate one aspect of the creative process by picking seemingly unrelated subjects, one we deem important, the other completely random, and then highlighting correlations between the two.
Example: Life vs. a car tire. Both involve some element of travel, and encounter obstacles large and small, which wear on the two in ways that eventually shape them into uniqueness that ultimately define them.
We get a symbol. Now to write a story about a tire. Better yet, just use it as one symbol in the story, and repeat the process to create a symbol repository.
Utilizing data mining techniques (specifically, pattern recognition and classification algorithms), the process could be automated to create such a symbol repository.
- Data sets that include metadata about their headers that link to additional data sets. That is, linked/nested data sets. “Definitions.”
- A standardized, public model for such data sets so others can contribute. Alternatively, some sort of system of conversion for already existing data sets found via a web crawler. Not a trivial task, though, so a standardized model and willing contributors would be a better immediate solution.
- A central repository of all compatible sets. A database of database references, basically.
- Optional: A lexer/grammar/syntax engine to spit out results in human readable format. That is, sentences describing found correlations. Also not a trivial task.
Note, I don’t mean to imply this would be a perfect simulation of “the creative process,” just one approach that could lead to some interesting results.