, and
,
the begin sentence, and begin paragraph
tags.
ASCII or binary (by default) file
containing a numerically sorted list of
Id n-gram n-tuples of numbers, corresponding to .id3gram.bin,
file the mapping of the word n-grams .id4gram.ascii etc.
relative to the vocabulary. Out of
vocabulary (OOV) words are mapped to
the number 0.
Binary file containing all the n-gram
Binary counts, together with discounting
language information and back-off weights. Can .binlm
model file be read by evallm and used to generate
word probabilities quickly.
ARPA language ASCII file containing the language
model file model probabilities in ARPA-standard .arpa
format.
ASCII file containing a list of
probabilities (one per line). The
Probability probabilities correspond the the
stream probability for each word in a specific.fprobs
text stream, with context-cues and OOVs
removed.
ASCII file containing a list of
Forced vocabulary words from which to enforce
back-off file back-off, together with either an 'i' .fblist
or an 'e' to indicate inclusive or
exclusive forced back-off respectively.
These files may all be written are read by all the tools in compressed or
uncompressed mode. Specifically, if a filename is given a .Z extension, then
it will be read from the specified file via a zcat pipe, or written via a
compress pipe. If a filename is given a .gz, it will be read from the
specified file via a gunzip pipe, or written via a gzip pipe. If either of
these compression schemes are to be used, then the relevant tools (ie zcat,
and compress or gzip) must be available on the system, and pointed to by the
path.
If a filename argument is given as - then it is assumed to represent either
the standard input, or standard output (according to context). Any file read
from the standard input is assumed to be uncompressed, and therefore, all
desired compression and decompression should take place in a pipe: zcat <
abc.Z | abc2xyz | compress > xyz.Z
----------------------------------------------------------------------------
The Tools
Note that in addition to the command line options mentioned, all the tools
also support -help and -version.
text2wfreq
Input : Text stream
Output : List of every word which occurred in the text, along with its
number of occurrences.
Notes : Uses a hash-table to provide an efficient method of counting word
occurrences. Output list is not sorted (due to "randomness" of the
hash-table), but can be easily sorted into the user's desired order by the
UNIX sort command. In any case, the output does not need to be sorted in
order to serve as input for wfreq2vocab.
Command Line Syntax:
text2wfreq [ -hash 1000000 ]
[ -verbosity 2 ]
< .text > .wfreq
Higher values for the -hash parameter require more memory, but can reduce
computation time.
wfreq2vocab
Input : A word unigram file, as produced by text2wfreq
Output : A vocabulary file.
Command Line Syntax:
wfreq2vocab [ -top 20000 | -gt 10]
[ -records 1000000 ]
[ -verbosity 2]
< .wfreq > .vocab
The -top parameter allows the user to specify the size of the vocabulary; if
the program is called with the command -top 20000, then the vocabulary will
consist of the most common 20,000 words.
The -gt parameter allows the user to specify the number of times that a word
must occur to be included in the vocabulary; if the program is called with
the command -gt 10, then the vocabulary will consist of all the words which
occurred more than 10 times.
If neither the -gt, nor the -top parameters are specified, then the program
runs with the default setting of taking the top 20,000 words.
The -records parameter allows the user to specify how many of the word and
count records to allocate memory for. If the number of words in the input
exceeds this number, then the program will fail, but a high number will
obviously result in a higher memory requirement.
text2wngram
Input : Text stream
Output : List of every word n-gram which occurred in the text, along with
its number of occurrences.
Command Line Syntax:
text2wngram [ -n 3 ]
[ -temp /usr/tmp/ ]
[ -chars n ]
[ -words m ]
[ -gzip | -compress ]
[ -verbosity 2 ]
< .text > .wngram
The maximum numbers of charactors and words that can be stored in the buffer
are given by the -chars and -words options. The default number of characters
and words are chosen so that the memory requirement of the program is
approximately that of STD_MEM, and the number of charactors is seven times
greater than the number of words.
The -temp option allows the user to specify where the program should store
its temporary files.
text2idngram
Input : Text stream, plus a vocabulary file.
Output : List of every id n-gram which occurred in the text, along with its
number of occurrences.
Notes : Maps each word in the text stream to a short integer as soon as it
has been read, thus enabling more n-grams to be stored and sorted in memory.
Command Line Syntax:
text2idngram -vocab .vocab
[ -buffer 100 ]
[ -temp /usr/tmp/ ]
[ -files 20 ]
[ -gzip | -compress ]
[ -n 3 ]
[ -write_ascii ]
[ -fof_size 10 ]
[ -verbosity 2 ]
< .text > .idngram
By default, the id n-gram file is written out as binary file, unless the
-write_ascii switch is used.
The size of the buffer which is used to store the n-grams can be specified
using the -buffer parameter. This value is in megabytes, and the default
value can be changed from 100 by changing the value of STD_MEM in the file
src/toolkit.h before compiling the toolkit.
The program will also report the frequency of frequency of n-grams, and the
corresponding recommended value for the -spec_num parameters of idngram2lm.
The -fof_size parameter allows the user to specify the length of this list.
A value of 0 will result in no list being displayed.
The -temp option allows the user to specify where the program should store
its temporary files.
In the case of really huge quantities of data, it may be the case that more
temporary files are generated than can be opened at one time by the filing
system. In this case, the temporary files will be merged in chunks, and the
-files parameter can be used to specify how many files are allowed to be
open at one time.
ngram2mgram
Input : Either a word n-gram file, or an id n-gram file.
Output : Either a word m-gram file, or an id m-gram file, where m < n.
Command Line Syntax:
ngram2mgram -n N -m M
[ -binary | -ascii | -words ]
< .ngram > .mgram
The -binary, -ascii, -words correspond to the format of the input and output
(Note that the output file will be in the same format as the input file).
-ascii and -binary denote id n-gram files, in ASCII and binary formats
respectively, and -words denotes a word n-gram file.
wngram2idngram
Input : Word n-gram file, plus a vocabulary file.
Output : List of every id n-gram which occurred in the text, along with its
number of occurrences, in either ASCII or binary format.
Note : For this program to be successful, it is important that the
vocabulary file is in alphabetical order. If you are using vocabularies
generated by the wfreq2vocab tool then this should not be an issue, as they
will already be alphabetically sorted.
Command Line Syntax:
wngram2idngram -vocab .vocab
[ -buffer 100 ]
[ -hash 200000 ]
[ -temp /usr/tmp/ ]
[ -files 20 ]
[ -gzip | -compress ]
[ -verbosity 2 ]
[ -n 3 ]
[ -write_ascii ]
< .wngram > .idngram
The size of the buffer which is used to store the n-grams can be specified
using the -buffer parameter. This value is in megabytes, and the default
value can be changed from 100 by changing the value of STD_MEM in the file
src/toolkit.h before compiling the toolkit.
The program will also report the frequency of frequency of n-grams, and the
corresponding recommended value for the -spec_num parameters of idngram2lm.
The -fof_size parameter allows the user to specify the length of this list.
A value of 0 will result in no list being displayed.
Higher values for the -hash parameter require more memory, but can reduce
computation time.
The -temp option allows the user to specify where the program should store
its temporary files.
The -files parameter is used to specify the number of files which can be
open at one time.
idngram2stats
Input : An id n-gram file (in either binary (by default) or ASCII (if
specified) mode).
Output : A list of the frequency-of-frequencies for each of the 2-grams, ...
, n-grams, which can enable the user to choose appropriate cut-offs, and to
specify appropriate memory requirements with the -spec_num option in
idngram2lm.
Command Line Syntax:
idngram2stats [ -n 3 ]
[ -fof_size 50 ]
[ -verbosity 2 ]
[ -ascii_input ]
< .idngram > .stats
mergeidngram
Input : A set of id n-gram files (in either binary (by default) or ASCII (if
specified) format - note that they should all be in the same format,
however).
Output : One id n-gram file (in either binary (by default) or ASCII (if
specified) format), containing the merged id n-grams from the input files.
Notes : This utility can also be used to convert id n-gram files between
ascii and binary formats.
Command Line Syntax:
mergeidngram [ -n 3 ]
[ -ascii_input ]
[ -ascii_output ]
.idngram_1 .idngram_2 ... .idngram_N > .idngram
idngram2lm
Input : An id n-gram file (in either binary (by default) or ASCII (if
specified) format), a vocabulary file, and (optionally) a context cues file.
Additional command line parameters will specify the cutoffs, the discounting
strategy and parameters, etc.
Output : A language model, in either binary format (to be read by evallm),
or in ARPA format.
Command Line Syntax:
idngram2lm -idngram .idngram
-vocab .vocab
-arpa .arpa | -binary .binlm
[ -context .ccs ]
[ -calc_mem | -buffer 100 | -spec_num y ... z ]
[ -vocab_type 1 ]
[ -oov_fraction 0.5 ]
[ -linear | -absolute | -good_turing | -witten_bell ]
[ -disc_ranges 1 7 7 ]
[ -cutoffs 0 ... 0 ]
[ -min_unicount 0 ]
[ -zeroton_fraction 1.0 ]
[ -ascii_input | -bin_input ]
[ -n 3 ]
[ -verbosity 2 ]
[ -four_byte_counts ]
[ -two_byte_bo_weights
[ -min_bo_weight -3.2 ] [ -max_bo_weight 2.5 ]
[ -out_of_range_bo_weights 10000 ] ]
The -context parameter allows the user to specify a file containing a list
of words within the vocabulary which will serve as context cues (for
example, markers which indicate the beginnings of sentences and paragraphs).
-calc_mem, -buffer and -spec_num x y ... z are options to dictate how it is
decided how much memory should be allocated for the n-gram counts data
structure. -calc_mem demands that the id n-gram file should be read twice,
so that we can accurately calculate the amount of memory required. -buffer
allows the user to specify an amount of memory to grab, and divides this
memory equally between the 2,3, ..., n-gram tables. -spec_num allows the
user to specify exactly how many 2-grams, 3-grams, ... , and n-grams will
need to be stored. The default is -buffer STD_MEM.
The toolkit provides for three types of vocabulary, which each handle
out-of-vocabulary (OOV) words in different ways, and which are specified
using the -vocab_type flag.
A closed vocabulary (-vocab_type 0) model does not make any provision for
OOVs. Any such words which appear in either the training or test data will
cause an error. This type of model might be used in a command/control
environment where the vocabulary is restricted to the number of commands
that the system understands, and we can therefore guarantee that no OOVs
will occur in the training or test data.
An open vocabulary model allows for OOVs to occur; out of vocabulary words
are all mapped to the same symbol. Two types of open vocabulary model are
implemented in the toolkit. The first type (-vocab_type 1) treats this
symbol the same way as any other word in the vocabulary. The second type
(-vocab_type 2) of open vocabulary model is to cover situations where no
OOVs occurred in the training data, but we wish to allow for the situation
where they could occur in the test data. This situation could occur, for
example, if we have a limited amount of training data, and we choose a
vocabulary which provides 100% coverage of the training set. In this case,
an arbitrary proportion of the discount probability mass (specified by the
-oov_fraction option) is reserved for OOV words.
The discounting strategy and its parameters are specified by the -linear,
-absolute, -good_turing and -witten_bell options. With Good Turing
discounting, one can also specify the range over which discounting occurs,
using the -disc_ranges option.
The user can specify the cutoffs for the 2-grams, 3-grams, ..., n-grams by
using the -cutoffs parameter. A cutoff of K means that > n-grams occurring K
or fewer times are discarded. If the parameter is omitted, then all the
cutoffs are set to zero.
The -zeroton_fraction option specifies that P(zeroton) (the unigram
probability assigned to a vocabulary word that did not occurred at all in
the training data) will be at least that fraction of P(singleton) (the
probability assigned to a vocabulary word that occurred exactly once in the
training data).
By default, the n-gram counts are stored in two bytes by use of a count
table (this allows the counts to exceed 65535, while keeping the data
structures used to store the model compact). However, if more than 65535
distinct counts need to be stored (very unlikely, unless constructing 4-gram
or higher language models using Good-Turing discounting), the
-four_byte_counts option will need to be used.
The floating point values of the back-off weights may be stored as two-byte
integers, by using the -two_byte_alphas switch. This will introduce slight
rounding errors, and so should only be used if memory is short. The
-min_alpha, -max_alpha and -out_of_range_alphas are parameters used by the
functions for using two-byte alphas. Their values should only be altered if
the program instructs it. For further details, see the comments in the
source file src/two_byte_alphas.c.
binlm2arpa
Input : A binary format language model, as generated by idngram2lm.
Output : An ARPA format language model.
Command Line Syntax:
binlm2arpa -binary .binlm
-arpa .arpa
[ -verbosity 2 ]
evallm
Input : A binary or ARPA format language model, as generated by idngram2lm.
In addition, one may also specify a text stream to be used to compute the
perplexity of the language model. The ARPA format language model does not
contain information as to which words are context cues, so if an ARPA format
lanaguage model is used, then a context cues file may be specified as well.
Output : The program can run in one of two modes.
* compute-PP - Output is the perplexity of the language model with
respect to the input text stream.
* validate - Output is confirmation or denial that the sum of the
probabilities of each of the words in the context supplied by the user
sums to one.
Command Line Syntax:
evallm [ -binary .binlm |
-arpa .arpa [ -context .ccs ] ]
Notes: evallm can receive and process commands interactively. When it is
run, it loads the language model specified at the command line, and waits
for instructions from the user. The user may specify one of the following
commands:
* perplexity
Computes the perplexity of a given text. May optionally specify words
from which to force back-off.
Syntax:
perplexity -text .text
[ -probs .fprobs ]
[ -oovs .oov_file ]
[ -annotate .annotation_file ]
[ -backoff_from_unk_inc | -backoff_from_unk_exc ]
[ -backoff_from_ccs_inc | -backoff_from_ccs_exc ]
[ -backoff_from_list .fblist ]
[ -include_unks ]
If the -probs parameter is specified, then each individual word
probability will be written out to the specified probability stream
file.
If the -oovs parameter is specified, then any out-of-vocabulary (OOV)
words which are encountered in the test set will be written out to the
specified file.
If the -annotate parameter is used, then an annotation file will be
created, containing information on the probability of each word in the
test set according to the language model, as well as the back-off class
for each event. The back-off classes can be interpreted as follows:
Assume we have a trigram language model, and are trying to predict P(C
| A B). Then back-off class "3" means that the trigram "A B C" is
contained in the model, and the probability was predicted based on that
trigram. "3-2" and "3x2" mean that the model backed-off and predicted
the probability based on the bigram "B C"; "3-2" means that the context
"A B" was found (so a back-off weight was applied), "3x2" means that
the context "A B" was not found.
To force back-off from all unknown words, use the -backoff_from_unk_inc
or -backoff_from_unk_exc flag (the difference being the difference
between inclusive or exclusive forced back-off). To force back-off from
all context-cues, use the -backoff_from_ccs_inc or
-backoff_from_ccs_inc flag. One can also specify a list of words from
which to back-off, by storing this list in a forced back-off list file
and using the -backoff_from_list switch.
-include_unks results in a perplexity calculation in which the
probability estimates for the unkown word are included.
* validate
Calculate the sum of the probabilities of all the words in the
vocabulary given the context specified by the user.
Syntax:
validate [ -backoff_from_unk_inc | -backoff_from_unk_exc ]
[ -backoff_from_ccs_inc | -backoff_from_ccs_exc ]
[ -backoff_from_list .fblist ]
word1 word2 ... word_(n-1)
Where n is the n in n-gram.
* help
Displays a help message.
Syntax:
help
* quit
Exits the program.
Syntax:
quit
Since the commands are read from standard input, a command file can be piped
into it directly, thus removing the need for the program to run
interactively:
echo "perplexity -text b.text" | evallm -binary a.binlm
interpolate
Input : Files containing probability streams, as generated by the -probs
option of the perplexity command of evallm. Alternatively these
probabilities could be generated from a seperate piece of code, which
assigns word probabilities according to some other language model, for
example a cache-based LM. This probability stream can then be linearly
interpolated with one from a standard n-gram model using this tool.
Output : An optimal set of interpolation weights for these probability
streams, and (optionally) a probability stream corresponding to the linear
combination of all the input streams, according to the optimal weights. The
optimal weights are calculated using the expectation maximisation (EM)
algorithm.
Command Line Syntax :
interpolate +[-] model1.fprobs +[-] model2.fprobs ...
[ -test_all | -test_first n | -test_last n | -cv ]
[ -tag .tags ]
[ -captions .captions ]
[ -in_lambdas .lambdas ]
[ -out_lambdas .lambdas ]
[ -stop_ratio 0.999 ]
[ -probs .fprobs ]
[ -max_probs 6000000 ]
The probability stream filenames are prefaced with a + (or a +- to indicate
that the weighting of that model should be fixed).
There are a range of options to determine which part of the data is used to
calculate the weights, and which is used to test them. One can test the
perplexity of the interpolated model based on all the data, using the
-test_all option, in which case a set of lambdas must also be specified with
the -lambda option (ie the lambdas are pre-specified, and not calculated by
the program). One can specify that the first or last n items are the test
set by use of the -test_first n or -test_last n options. Or one can perform
two-way cross validation using the -cv option. If none of these are
specified then the whole of the data is used for weight estimation.
By default, the initial interpolation weights are fixed as
1/number_of_models, but alternative values can be stored in a file and used
via the -in_lambdas option.
The -probs switch allows the user to specify a filename in which to store
the combined probability stream. The optimal lambdas can also be stored in a
file by use of the -out_lambdas command.
The program stops when the ratio of the test-set perplexity between two
successive iterations is above the value specified in the -stop_ratio
option.
The data can be partitioned into different classes (with optimisation being
performed seperately on each class) using the -tags parameter. The tags file
will contain an integer for each item in the probability streams
corresponding to the class that the item belongs to. A file specified using
the -captions option will allow the user to attach names to each of the
classes. There should be one line in the captions file for each tag, with
each line corresponding to the name of the tag.
The amount of memory allocated to store the probability streams is dictated
by the -max_probs option, which indicates the maximum number of
probabilities allowed in one stream.
Note: For an example use and output of a previous version of this program
(with slightly different syntax), see Appendix B of R. Rosenfeld Adaptive
Statistical Language Modeling: A Statistical Approach PhD Thesis, School of
Computer Science, Carnegie Mellon University, April 1994. Published as
Techical Report CMU-CS-94-138
----------------------------------------------------------------------------
Typical Usage
[Simplified toolkit framework - 8KB]
Given a large corpus of text in a file a.text, but no specified vocabulary
* Compute the word unigram counts
cat a.text | text2wfreq > a.wfreq
* Convert the word unigram counts into a vocabulary consisting of the
20,000 most common words
cat a.wfreq | wfreq2vocab -top 20000 > a.vocab
* Generate a binary id 3-gram of the training text, based on this
vocabulary
cat a.text | text2idngram -vocab a.vocab > a.idngram
* Convert the idngram into a binary format language model
idngram2lm -idngram a.idngram -vocab a.vocab -binary a.binlm
* Compute the perplexity of the language model, with respect to some test
text b.text
evallm -binary a.binlm
Reading in language model from file a.binlm
Done.
evallm : perplexity -text b.text
Computing perplexity of the language model with respect
to the text b.text
Perplexity = 128.15, Entropy = 7.00 bits
Computation based on 8842804 words.
Number of 3-grams hit = 6806674 (76.97%)
Number of 2-grams hit = 1766798 (19.98%)
Number of 1-grams hit = 269332 (3.05%)
1218322 OOVs (12.11%) and 576763 context cues were removed from the
calculation.
evallm : quit
Alternatively, some of these processes can be piped together:
cat a.text | text2wfreq | wfreq2vocab -top 20000 > a.vocab
cat a.text | text2idngram -vocab a.vocab | \
idngram2lm -vocab a.vocab -idngram - \
-binary a.binlm -spec_num 5000000 15000000
echo "perplexity -text b.text" | evallm -binary a.binlm
----------------------------------------------------------------------------
Discounting Strategies
Discounting is the process of replacing the original counts with modified
counts so as to redistribute the probability mass from the more commonly
observed events to the less frequent and unseen events. If the actual number
of occurrences of an event E (such as a bigram or trigram occurrence) is
c(E), then the modified count is d(c(E))c(E), where d(c(E)) is known as the
discount ratio.
Good Turing discounting
Good Turing discounting defines d(r) = (r+1)n(r+1) / rn(r) where n(r) is the
number of events which occur r times.
The discounting is only applied to counts which occur fewer than K times,
where typically K is chosen to be around 7. This is the "discounting range"
which is specified using the -disc_ranges parameter of the idngram2lm
program.
For further details see "Estimation of Probabilities from Sparse Data for
the Language Model Component of a Speech Recognizer", Slava M. Katz, in
"IEEE Transactions on Acoustics, Speech and Signal Processing", volume
ASSP-35, pages 400-401, March 1987.
Witten Bell discounting
The discounting scheme which we refer to here as "Witten Bell discounting"
is that which is referred to as type C in "The Zero-Frequency Problem:
Estimating the Probabilities of Novel Events in Adaptive Text Compression",
Ian H. Witten and Timothy C. Bell, in "IEEE Transactions on Information
Theory, Vol 37, No. 4, July 1991".
The discounting ratio is not dependent on the event's count, but on t, the
number of types which followed the particular context. It defines d(r,t) =
n/(n + t), where n is the size of the training set in words. This is
equivalent to setting P(w | h) = c / (n + t) (where w is a word, h is the
history and c is the number of occurrences of w in the context h), for
events that have been seen, and P(w | h) = t / (n + t) for unseen events.
Absolute discounting
Absolute discounting defines d(r) = (r-b)/r. Typically b=n(1)/(n(1)+2n(2)).
The discounting is applied to all counts.
This is, of course, equivalent to simply subtracting the constant b from
each count.
Linear discounting
Linear discounting defines d(r) = 1 - (n(1)/C), where C is the total number
of events. The discounting is applied to all counts.
For further details of both linear and absolute discounting, see "On
structuring probabilistic dependencies in stochastic language modeling", H.
Ney, U. Essen and R. Kneser in "Computer Speech and Language", volume 8(1),
pages 1-28, 1994.
----------------------------------------------------------------------------
Up-to-date Information
The latest news on updates, bug fixes etc. can be found here.
----------------------------------------------------------------------------
Feedback
Any comments, questions or bug reports concerning the toolkit should be
addressed to Philip Clarkson.
----------------------------------------------------------------------------
Philip Clarkson - prc14@eng.cam.ac.uk
was required to be in the vocabulary. In
version 2, one may have any number of context cues (or none at all), and
they may be represented by any symbols one chooses. The context cues are a
subset of the vocabulary, and are specified in a context cue file.
In order to produce the same behaviour from version 2 as from version 1, the
context cues file should contain the following lines:
B (where
is a context cue indicating a sentence boundary), then instead of
predicting the probability of B based on the full context (P(B | A )), we
may wish to disregard the information before the sentence boundary.
Therefore we might want to back-off to the bigram distribution P(B | )
(inclusive forced back-off) or even to the unigram distribution P(B)
(exclusive forced back-off). Version 2 supports both types of forced
back-off for arbitrary vocabulary items.
The evallm program allows the user to specify either inclusive or exclusive
forced back-off, as well as a list of words from which to enforce back-off.
Installing the Toolkit
For "big-endian" machines (eg those running HP-UX, IRIX, SunOS, Solaris) the
installation procedure is simply to change into the src/ directory and type
make install
The executables will then be copied into the bin/ directory, and the library
file SLM2.a will be copied into the lib/ directory. For "little-endian"
machines (eg those running Ultrix, Linux) the variable BYTESWAP_FLAG will
need to be set in the Makefile. This can be done by editing src/Makefile
directly, so that the line
#BYTESWAP_FLAG = -DSLM_SWAP_BYTES
is changed to
BYTESWAP_FLAG = -DSLM_SWAP_BYTES
Then the program can be installed as before.
If you are unsure of the "endian-ness" of your machine, then the shell
script endian.sh should be able to provide some assistance.
In case of problems, then more information can be found by examining
src/Makefile.
Before building the executables, it might be worth adjusting the value of
STD_MEM in the file src/toolkit.h. This value controls the default amount of
memory (in MB) that the programs will attempt to assign for the large
buffers used by some of the programs (this value can, of course, be
overridden at the command line). The result is that the final process sizes
will be a few MB bigger than this value. The more memory that can be
grabbed, the faster the programs will run. The default value is 100, but if
the machines which the tools will be run on contain less, or much more
memory than this, then this value should be adjusted to reflect this.
----------------------------------------------------------------------------
Terminology and File Formats
Name Description Typical file
extension
An ASCII file containing text. It may
Text stream or may not have markers to indicate .text
context cues, and white space can be
used freely.
An ASCII file containing a list of
Word words, and the number of times that
frequency they occurred. This list is not sorted;.wfreq
file it will generally be used as the input
to wfreq2vocab, which does not require
sorted input.
Word n-gram ASCII file containing an alphabetically
file sorted list of n-tuples of words, along.w3gram, .w4gram etc.
with the number of occurrences
ASCII file containing a list of
.vocab.20K,
Vocabulary vocabulary words. Comments may also be .vocab.60K etc.,
file included - any line beginning ## is depending on the size
considered a comment. The vocabulary is
limited in size to 65535 words. of the vocabulary.
ASCII file containing the list of words
which are to be considered "context
cues". These are words which provide
Context cues useful context information for the
file n-grams, but which are not to be .ccs
predicted by the language model.
Typical examples would be and