-
Notifications
You must be signed in to change notification settings - Fork 21
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Co-authored Procedural Novel #83
Comments
Very OuLiPo-ish. |
So I've moved away from picking random sentences and instead am using I use the above multiple-choice program but what I'm doing instead of random sentences is going through each line of Pride and Prejudice and asking it for the ten most similar lines in the corpus. If I were to carelessly pick the most similar line each time, here's what the first 20 lines of the novel would look like:
Now here's the "edited" version I came up with for those first 20 lines:
Spicy! |
New algorithm: instead of going line-by-line through Pride and Prejudice, we branch through all the books we have available. Like this:
Here's the result of one run:
|
Here's another run of the same jumping-around algorithm. It's definitely less coherent but it does this neat thing where it sort of jumps between styles.
I'm suspecting this is TOO random -- I may have it do mostly the "follow a single book and create analogies from that" thing for a while and then every now and then have it skip. Just to keep things interestingl |
Here is an example where it mostly sticks to a bunch of passages in a row but then branches out every now and then to a different passage:
|
And another one with sometimes-random branching, this time with semi-random newlines (there's always a new line on new dialogue starting with
|
Here's a complete book, though it's not quite 50,000 words (it's more like 48k). http://tinysubversions.com/nanogenmo/2015/harpooneers/ NOT my official entry for this since it doesn't meet the length requirement. |
Err, updated with the source code. I'm sorry: https://github.com/dariusk/harpooneers |
@hugovk I was considering marking this as complete since it comes awfully close and does include the source code... what do you think? |
If you include the source code (but not the corpora) in an appendix in the back of the final, printed volume it's almost 52k. Congratulations, have a Completed label! |
Thought experiment: you and I are playing a game. I write ten opening sentences for a novel, and you pick the one you like best and let me know; that becomes the opener. Then I write ten second sentences for the novel. You pick what you like best and let me know. Et cetera.
Who wrote the book? I wrote literally every word, but you dictated nearly the entire form of the novel.
I plan to act as sentence-by-sentence editor for an algorithm (or set of algorithms) where I review something like 5,000 multiple choice questions and hand pick each sentence of the novel.
Right now I'm using an extremely naive function that just pulls random sentences from Gutenberg using gutencorpus. Here's ten sentences, derived from 10x10 sentences that the algorithm gave me:
I picked the order of the sentences and I do think it hangs together with more coherence than quick experiment like this usually does. My plan is to use
doc2vec
or similar, with maybe a few other tricks, and have the algorithm take into account at least the very last sentence I gave it to give me a more coherent set of options (with this pure random sentence thing, usually only 1 or 2 out of 10 options makes any sense at all).The text was updated successfully, but these errors were encountered: