search

tulul / tululbot

Telegram Bot for Tululness
Apache License 2.0
8 stars 6 forks source link

Scoring for Tulul Quotes Single Random Retrieval #36

Open araishikeiwai opened 8 years ago

araishikeiwai commented 8 years ago

Quotes should be scored in a way such that no quote appears too often or never appears at all

wazaundtechnik commented 8 years ago

Hi rick. Does it mean we are not allowed to use Gaussian Random anymore?

kmkurn commented 8 years ago

Any idea what the probability distribution should look like? Right now we're using Python's built-in random.choice() function and from the docs it says the distribution is uniform.

araishikeiwai commented 8 years ago

quotes which appear often should have lower priority. so each quote has to be weighed. when a quote is picked, add 1 to the weigh. higher weigh = lower priority.

araishikeiwai commented 8 years ago

look: http://www.evanmiller.org/how-not-to-sort-by-average-rating.html

not sure how to utilise since we only have 1 variable