Actually, the work of Salton's that I'm familier with is
the exact antithesis of machine learning. From what I've
read, he favors algorithmic versus semantic styles of
classification. That is, he counts term frequencies and
compares documents on that rather than try to understand
semantics. (Sorry if I'm reading too much into the phrase
"machine learning techniques". I'm not completely sure myself
what that might mean.)
Anyway, a good start for Salton's work is in the February
1994 issue of Communications of the ACM (37(2)).
PF