Thanks for the fast reply!

> Great question and great idea! The way we get images from Flickr in
> the TagMaps application, right now, ...

> a. Get the tags for each region/zoom level using the TagMaps API
> http://developer.yahoo.com/yrb/tagmaps/
> b. Get (however many) images for each tag on the map using a
> text-and-bounding-box query from Flickr:
> http://flickr.com/services/api/flickr.photos.search.html
> c. Sort the photos however way you like :)

I was hoping of jumping from (a) to (c) with the specific images that
TagMaps has recognized as relevant to that tag. At least from
appearances in the flash app, it seems that the number (and quality)
of images that appear for a specific tag seem to have been refined
from what I might find by doing a direct tag search via the Flickr
interface. Is this true or are my expectations just lower because I'm
used to unassisted browsing by tags?

In another direction entirely, can you tell me if you do any
correlation between the tags suggested by TagMaps and the tags that
Flickr seems to statistically predict with it's "getRelated" function?
(sorry I would provide the exact url for clarity, but it seems the
interface is down) By this I'm wondering if you leverage the tags
that you've found from one geographical area in TagMaps versus the
tags that may be statistically related to give the smaller list that
appears when I hover over an area. If not, how _do_ you create that
smaller list of suggested tags ;) ?

Thanks again and I look forward to working with this great data.

Eric