Violet, Queen Of Everything (violetiae) wrote in matchingrebequa,
Violet, Queen Of Everything

  • Mood:

Group Huddle: Help With Matching Ideas

Since the match feature isn't working, we may need to come up with another way for folks to figure whether they match Re'bequa.

I was thinking percentages might be good to sort by until the match thing gets fixed. But what would be a good percentage? Twenty percent in common with Re'bequa (i.e., 1 out of 5 interests)? 25% (1 out of 4)? 30% (about 1 in 3)?, or what?

What would happen for people who don't have 150 interests listed. What if they only have, say, 30 interests listed and 6 of those match Re'bequa's? Would that count as a 20% match?

Does anyone have any other ideas for ways we can give people to tell if they match Re'bequa? It doesn't even have to be that rigid. I don't have a problem with people joining just because they find it funny. It's the spirit of the thing. But since the entire concept of the community is matching Re'bequa, it would be nice if people could have some temporary guidelines for alternate ways "matching" might be determined while the feature is down.

I sure do wish LJ would get that feature fixed!
  • Post a new comment


    Anonymous comments are disabled in this journal

    default userpic

    Your IP address will be recorded 

Perhaps someone (far more clever in such matters than I) could whip up an off-site script to do the match? We can hard-code Re'bequa's interests, since they won't be changing; the script would just have to get the user's interests. I've seen a number of journal spamming thingies that do this sort of stuff, so there must be a way to pull a list of interests off the LJ servers by username.

If someone can come up with a nice CGI/PHP/Perl/whatever thing, I'll host it.
As for the mechanics of the match: since Re'bequa's sitting solid at 150 interests, I think it would make most sense to display what percentage of the user's interests match Re'bequa's (I'm getting a little sick of typing that extra apostrophe right now, truth be told). That way, it wouldn't matter whether someone had few or many interests listed, they could still be a high or low match.
the problem or rather the intrigue of the original matching feature was that it didn't just use percentages, it had some sort of weight factor. I was always under the impression that LJ was implementing an algorithm that took in consideration the total number a single interest was mentioned in the entire database, and gave it extra weight in the overall matchup the fewer times it was mentioned. I think rebequa's special sauce was/is that an extremely high (out of proportion for the average LJ user) percentage of her interests have a low mention, or at least enough of her interests are low *enough* that she garners a special weight of her own in the indexing system. that is to say, she's an anomaly in the system, an uber-meta-interest, if you will. mr. anderson, yadayadayada, blade runner, yadayadayada, my point being, rare that it is that I have one, that duplicating the methodology would be difficult, but the right person could come up with a creative solution that approximates it. what could be interesting and would garner more attention to the community, (if we desire to be more inclusive and less rebequa elite worship oriented) is if we could turn it into one of those memes that people can post in their journals. you know like a quizilla thing. I'm just sayin.

ooohhh if only i had more free time "which one of re'bequa's interests are you?"


"which one of re'bequa's adoring fans are you?"
Didn't the match thing go beyond just "interests" I thought there was something about "communities" both people belonged to, "friends" in common, etc. adding a little extra weight to the rankings.