Ellibereth show us your meta sheets

[Ellibereth (Elli)]

did you train your own model or used something pretrained

[Cactus]

Interesting! What will you be using?

Probably just text (words and length), post times, and relevant vote info (time between, position etc.)

Sorry, I meant what languages/algorithms/models/etc. Would you be sticking with JS?

[GTacc (Espresso GTacc)]

Yeah this is where I stop reading thread

[Cactus]

did you train your own model or used something pretrained

Let’s call it an ad hoc ensemble method, using three APIs. I’m not sure what exactly they have under the hood, though one of them appeared to be AFINN junk so I underweighted it.

[Cactus]

Confirmed for AFINN junk.

Every student’s gotta make one, apparently.

[Ellibereth (Elli)]

Sorry, I meant what languages/algorithms/models/etc. Would you be sticking with JS?

nah, prolly python.

v1’s going to be super naive e.g. search string -> see how often each alignment used by context -> filter by player, show a player’s post frequency and length are per alignment etc.
No learning just brute force.

Probably eventually makes sense to look at stuff with https://nlp.stanford.edu/projects/glove/ etc. but that’ll be pretty far from now. Need to get automation and tracking stuff done before we can do most of this.

If you have any suggestions for what should be running behind the hood lemme know. Haven’t kept up with nlp stuff at all post graduation.

[M2H (gamethrowing prodigy)]

Yeah this is where I stop reading thread

[Cactus]

Sorry, I meant what languages/algorithms/models/etc. Would you be sticking with JS?

nah, prolly python.

v1’s going to be super naive e.g. search string -> see how often each alignment used by context -> filter by player, show a player’s post frequency and length are per alignment etc.
No learning just brute force.

Probably eventually makes sense to look at stuff with https://nlp.stanford.edu/projects/glove/ etc. but that’ll be pretty far from now. Need to get automation and tracking stuff done before we can do most of this.

If you have any suggestions for what should be running behind the hood lemme know. Haven’t kept up with nlp stuff at all post graduation.

Thumbs up for Python! That would be my preference as well.

Have you looked at some of the NLP APIs out there? You can do quite a bit of free traffic with Google, IBM, etc. Obviously there would be some privacy concerns, but maybe that’s not an issue.

[Ellibereth (Elli)]

yeah I’ll take a look

We want to make it so that the site’s games are an easy to use dataset for people who want to play around with it (technical or otherwise) so I’m glad there’s enthusiasm for this sort of thing.

[chesskid3]

In my experience semen analysis doesn’t test that well for stds.

[Cactus]

In my experience semen analysis doesn’t test that well for stds.

“analysis” is just sitting there…

[Ellibereth (Elli)]

i couldn’t remember how to spell analysis for the life of me until some pointed out that I should just remember it was anal. Never got it wrong since.

[Tommy2Hands (フロー団 // 果てし無い)]

MU about to look rly clowny…

[Gamma_Emerald]

Interesting stuff tbh

[tn5421 (Neopolitan)]

nah, prolly python.

v1’s going to be super naive e.g. search string -> see how often each alignment used by context -> filter by player, show a player’s post frequency and length are per alignment etc.
No learning just brute force.

Probably eventually makes sense to look at stuff with https://nlp.stanford.edu/projects/glove/ etc. but that’ll be pretty far from now. Need to get automation and tracking stuff done before we can do most of this.

If you have any suggestions for what should be running behind the hood lemme know. Haven’t kept up with nlp stuff at all post graduation.

Sounds interesting but it also has the potential to have an IIoA effect on some of us in situations where we can’t rely on it, if we get used to it first.

[Key]

that has the potential to make our site scum game literally the best on the internet which is super hype.

Yeah developing tools for people to study and get better is something I’m personally super excited for in the future