@alex It scans all public posts in the window for mentions, hashtags, and non-dictionary words. Then, the N ones that occur the most during that window are selected and the rest are dropped (these N ones are used for the "top N" and the graphs). Lastly, for each hour, those N ones are counted for the graphs.
@alex It's a two-part thing, by the way. The server dumps a lot of post data (nothing private and a very limited subset of each post's data) and the client does the heavy munging.
@alex If you remind me tomorrow, I could send you what I put together for the datetime index. Yes, we can't always trust what comes from other servers, but everybody complains about months-old posts showing up.
@tk@alex Wouldn't it be better to rank them by sorting by how many different users are using the hashtags? Then stuff that more people are talking about would trend higher than posts from bots
@realcaseyrollins@alex I don’t really plan on messing with this anytime soon because I’m not in a great mental state for free time programming. The code is here if anyone wants to put up an MR.
@realcaseyrollins @alex I don’t really plan on messing with this anytime soon because I’m not in a great mental state for free time programming. The code is here if anyone wants to put up an MR.