Any one knows what kind of algorithm Google might have used in this feature or any public papers/ patents by Google related to this feature?
My interest in this feature is due to a paper which i wrote (and working on) in International Semantic Web Conference this year [ http://data.semanticweb.org/conference/iswc/2009/paper/poster_demo/130 ] and which is based on a paper by Common Sense Computing group at MIT Media Lab [ http://xnet.media.mit.edu/analogyspace ] .
Akshay Bhat