I'll po YOUR lite! (algeh) wrote in lj_dev,
I'll po YOUR lite!
algeh
lj_dev

Interest matching: how small would the db hit have to be to re-implement?

I've been thinking (although probably not as hard as I should be, as I'm sure y'all will soon tell me) and I was wondering if there would be interest in having a tool that supplied something somewhere in between what we used to have with interest matching (which, understandably, was way too big of a db hit) and what we have now, which is simply lists of who shares a single interest.

I know when I used the old feature, I sure didn't need a list of everybody, I used it as a tool to try to get a person or two who was worth reading. Therefore, it seems to me that there might be user interest in having an incomplete search using some kind of interest matching, that would return a few likely journals without searching through everything. This wouldn't be useful for finding a specific, known journal the way a real search might(not that interest matching was really good for that to start with), but it would be a way to suggest journals that the user might like.

For example, would it be too much of a strain to highlight users who shared one specific interest (selected by the user) and any of their other interests, or who simply shared 2 specific interests?

I know I should poke at the code for a while and ask more concrete questions, but it basically comes down to me needing to know what level of use of resources would be allowed for such a thing, and then looking at the code to figure out how/if a search of some kind could be added that was that small of strain (maybe something that stopped after finding x number of users, maybe it simply isn't possible, etc).

So my basic question is, is there a consensus for how little of a strain such a search would be allowed to be before it would be reasonable to implement here? Is it just not worth it to have something that only returns an incomplete list? (Also, is someone else already working on this and should I therefore butt out?) Sorry to bother people if this is annoying. I like trying to figure this kind of stuff out, but I don't feel like trying to come up with something lj specific unless there is some small, vague chance of it being implemented.

(Edited to add paragraphs. Sorry about that. It still may be a dumb question, but at least now it's easier to read.)
Subscribe
  • Post a new comment

    Error

    Anonymous comments are disabled in this journal

    default userpic

    Your reply will be screened

    Your IP address will be recorded 

  • 11 comments