Better Search = More Data
I ran across this interesting post yesterday on how adding the right kind of additional data to a dataset did more to improve its searchability and relevance than creating a more complex algorithm to search it. Worth a read for those search-geeks out there like myself.
We have daily conversations here about how to improve the relevance of WorldCat's search and find ourselves tweaking the algorithm almost monthly, but it's amazing how much more relevant the results became (back in the early days of '06) just by adding the count of libraries that hold an item into the "data pool." This gets me excited to start folding user-contributed data (tags, reviews, what-have-you) into the mix as it makes sense and as critical mass builds. The benefits to discoverability could be tremendous.