Decisions, decisions
We're in the pre-beta crunch for Tinfinger, and I'm wrestling with several questions that will probably not be answered until after the beta is over. Namely:
- size of the reading lists: try to index as much as possible (Topix.net) or aim for quality over quantity (Memeorandum)?
- tinscore (the working name for our ranking algorithm): not just how to weight the various variables, but what operands to use... can it be done just by arithmetic or is calculus necessary?
- size of the result set: try to find as many people as possible in each story, or only record the most important X number of people in the story, or only those who reach Y tinscore?
- structure of human metadata: use keywords, i.e. normal English language structure, or use tags comprised of spaceless keyphrases?
- disambiguation, i.e. where two people in the same category have the same first and last name... God, that's a pain in the arse. Not to mention middle initials, titles, junior/senior, Roman numerals...
- size of the reading lists: try to index as much as possible (Topix.net) or aim for quality over quantity (Memeorandum)?
- tinscore (the working name for our ranking algorithm): not just how to weight the various variables, but what operands to use... can it be done just by arithmetic or is calculus necessary?
- size of the result set: try to find as many people as possible in each story, or only record the most important X number of people in the story, or only those who reach Y tinscore?
- structure of human metadata: use keywords, i.e. normal English language structure, or use tags comprised of spaceless keyphrases?
- disambiguation, i.e. where two people in the same category have the same first and last name... God, that's a pain in the arse. Not to mention middle initials, titles, junior/senior, Roman numerals...
posted by Paul Montgomery | 12:41 pm
8 Comments:
"size of the reading lists: try to index as much as possible (Topix.net) or aim for quality over quantity (Memeorandum)?"
I'd vote for quality anyday. There is already enough aggregating sites that follow the fill the page with mostly crap. Quality will always win out - may take longer but the readership is always more faithful IMO
all the best with tinfinger! here's to getting it away from stealth mode ;)
Good point Steven. I'm not a fan of those river-like front pages like Digg or del.icio.us. I guess I'm trying to have it both ways, which is more difficult than either approach on its own.
Thanks oreos!
Are you doing any live testing and research with regular web users of the alternatives you just mentioned that could benefit from user input? Will mean that you won't have to be wrestling in your mind so much with some of those questions.
Yes Anthony, that's partly what the beta's for. All the different equations and ranking scores and whatnot will be displayed as options so beta participants can pick which combinations they like the best.
"Yes Anthony, that's partly what the beta's for"
Did some-one mention beta? :) -- and how would one be able to get in on this little party?
just incase :) codenut at winextra dot com will look forward to any possible party invitations.
Thanks, just checking ;)
Post a Comment
<< Home