Quote:
Originally posted by SomeCreep
Lots of words. Reading required.
Summary? Cliff notes? 
|
S.sub.i =0, if f.sub.i >f.sub.max ;
S.sub.i =log(.sub.max /(f.sub.i -T.sup.2 +T)), if T.sup.2 <f.sub.i <f.sub.max ;
S.sub.i =log(f.sub.max /T), if T<f.sub.i <T.sup.2 ; and
S.sub.i =.epsilon.+((f.sub.i /T)(log(f.sub.max /T)-.beta.)), if f.sub.i .ltoreq.T,
where .epsilon. and T are user-definable values.
A.sub.fi ?j! that is a function of the word's dictionary salience and its associated tree scores (e.g.,A.sub.fi ?j!=((D.sub.j (f(A.sub.i ?j!))) .SIGMA.A.sub.i ?j!)). For example, A.sub.fi =((D.sub.j (((1/p).SIGMA..phi..sub.m ?j!)(.SIGMA.A.sub.i ?j!)), where .phi..sub.m ?j!=((n?j !)(((1+((-1) p))/2)-((1/max(A.sub.i ?j!))(.SIGMA.((.SIGMA.(-A.sub.i ?j!)A.sub.i ?j!) (x-1))))))).