Scoble asks search giants for “access to the variables.” In other words, let users play with the guts of the engine, so they can tune the results to their liking. I hope the folks at MSFT are listening to one of their own. This kind of transparency, while problematic from a spam standpoint, would be most excellent.
Bravo…
Scoble asks search giants for "access to the variables." In other words, let users play with the guts of the engine, so they can tune the results to their liking. I hope the folks at MSFT are listening to one of their own. This kind of transparency, while problematic from…
The prototype for this is Yuntis, 3 years in development at Stony Brook University.
Here is the demo:
http://yuntis.ecsl.cs.sunysb.edu/
The site indicates that source code is available through a BSD-style license, but the download link is disabled. (You can try maxim@cs.stonybrook.edu for clarification.)
Combined with Nutch, should comprise a solid base for the Linux-ization of basic text search.
This kind of thing could arguably help fight spam, as there would not longer be a single formula to optimize against, but rather a family of formulae. Developing pages which score highly against all parameter variations should be more difficult.
It’s not really problematic from spam standpoint. If access to certain variety of variables can be controlled, and moreover, can be saved and retrieved, it would allow for greater modifications by people who know what they’re doing.
For example, you would be able to download the settings file that your friend created, that would give additional boost to magazine articles, media sites and research publications. With proliferation of such settings files you’d be able to choose the ones that are the most useful for you.
What Alex said, but further: if I have a sets of saved values for search weighting (off the top of my head, say settings for general news, technology news, code, humor and individual people), there is a serious fragmentation of targets for which to optimize. People can happily share ‘search profiles’. Mine will likely be weighted differently than a lot of other people’s. There is strength in genetic diversity, even when what you’re talking about isn’t strictly genetic.
It is also different than RSS, which is a natural sounding analogue, but that is the topic for a different rant.
-j
Gary keeps the discussion going on this topic here…