In a previous paper the notion of 'using the Amazon metric to construct an image database based on what people do, not what they say' was introduced (see [1]). In that paper we described a case study setting where 20 ...