
This folder contains SRILM language model files and exported files
from the SortedProbabilityAccessor class.


Naming Conventions

The filename

  europarl_2-to-10_stripped.es-en.en.lm

indicates that the file contains the SRILM language model file with
unigrams and bigrams for the English half of the Spanish-to-English
Europarl translation corpus.

The filename

  europarl_2-to-10_stripped.es-en.en.ntp.sbp

indicates that the file contains the exported contents of a
SortedProbabilityAccessor object that holds the sorted LM
probabilities for the English half of the Spanish-to-English
Europarl corpus, and that the probabilities have been sorted
by second-word-in-the-bigram and then first-word ("ntp", or
"next to previous"), or sorted by first-word and then second-word
("ptn", or "previous to next"). The "sbp" is for "sorted bigram
probability".
