18.05.2015 Views

SRI-LM toolkit

SRI-LM toolkit

SRI-LM toolkit

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Witten-Bell Discounting<br />

• T/(N+T) gives the total “probability of unseen N-<br />

grams”, we need to divide this up among all the<br />

zero N-grams<br />

• We could just choose to divide it equally<br />

Z<br />

=<br />

i:<br />

∑<br />

c i<br />

= 0<br />

1<br />

Z is the total number of<br />

N-grams with count zero<br />

p * T<br />

= i Z(<br />

N + T<br />

)<br />

16

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!