Karl Shah-Jenner <shahjen@iinet.net.au> writes: > Some time back someone posed the question about how google image searching > works - did we ever uncover their secret? I don't think there's any great secret. They search (or more precisely *index*) the text around images, and (I suspect) the image file name, plus almost certainly the 'alt' and 'title' texts. The whole thing is fairly hit and miss, but (like machine translation) it's almost infinitely cheap, so the results are useful at least some of the time. Recently I found one of my pages featuring in the top row in a google image search for "Escamillo" - these things change around, and the very first picture is of a cat! Again, an image search for "Sano" finds a large number of files called things like sano.jpg, but (at least on the first two pages) not a whisper of any of my photos of Sano - whereas putting "sano photos" into regular google I come up number one. But I think they are gradually improving. Brian Chandler ---------------- geo://Sano.Japan.Planet_3 http://imaginatorium.org/
