Thanks for the question. The geo docs are being updated at the moment and we've noticed various bits of them have gone missing, including search. We're looking into why that is at the moment.
You can directly access the information on the method by visiting: http://dev.twitter.com/doc/get/geo/search You can use max_results to limit the number of places which are matched - just know we have a known issue at the moment where the API sometimes returns 0 results when max_results=1. You can also include the query=Staples%20Center to do a place name search In terms of relevance I assume you mean relevance based on the accuracy of the name to what you desire. The API does a very simple AND search for each word in the query you pass it. This means you would need to post process the results to find what you consider the most relevant one. Your simple algorithm description sounds like an approach that would work. Hope that helps, Matt On Fri, Jun 25, 2010 at 12:09 PM, ELB <ebrit...@gmail.com> wrote: > > I notice that in the example of the Staples Center in Los Angeles, the > best choice for "Staples Center" is the second result. We would like > to automate our script; do you have any suggestions for determining in > code which of the 5 responses is best? Meaning, for a given venue, > Twitter has multiple Place IDS - how can we automate it to select the > most relevant one? > > Just *conceivably* some simple algorithm like (a) ruling out results > that do not have an exact match somewhere inside the result, and (b) > then choosing the result that has the *fewest* extra characters > outside the match, might give good results much of the time. That is > one idea - any others? > > > also, as an aside, we can't find documentation for this geo/search > API. Do you know if documentation has been put up yet? > Specifically, though, is there a way to control the number of results > - say, to provide 10 results rather than just 5? > > -- Matt Harris Developer Advocate, Twitter http://twitter.com/themattharris