View on
MetaCPAN is shutting down
For details read Perl NOC. After June 25th this page will redirect to
Bridget McInnes > Text-NSP-1.21 >


Annotate this POD


Open  0
View/Report Bugs
Source   Latest Release: Text-NSP-1.27

NAME ^ - Divide huge text into pieces and run separately on each (and then combine)


Runs efficiently on a huge data.



Required Arguments:


Input to should be a -

1. Single plain text file


item 2. Single flat directory containing multiple plain text files


3. List of multiple plain text files


A complete path to a writable directory to which can write all intermediate and final output files. If DESTINATION does not exist, a new directory is created, otherwise, the current directory is simply used for writing the output files.

NOTE: If DESTINATION already exists and if the names of some of the existing files in DESTINATION clash with the names of the output files created by huge-count, these files will be over-written w/o prompting user.

Optional Arguments:

--split P

This option should be specified when SOURCE is a single plain file. huge-count will divide the given SOURCE file into P (approximately) equal parts, will run separately on each part and will then recombine the bigram counts from all these intermediate result files into a single bigram output that shows bigram counts in SOURCE.

If SOURCE file contains M lines, each part created with --split P will contain approximately M/P lines. Value of P should be chosen such that can be efficiently run on any part containing M/P lines from SOURCE. As #words/line differ from files to files, it is recommended that P should be large enough so that each part will contain at most million words in total.


Specify a file containing Perl regular expressions that define the tokenization scheme for counting. This will be provided to's --token option.

--nontoken NOTOKENFILE

Specify a file containing Perl regular expressions of non-token sequences that are removed prior to tokenization. This will be provided to the's --nontoken option.


Specify a file of Perl regex/s containing the list of stop words to be omitted from the output BIGRAMS. Stop list can be used in two modes -

AND mode declared with '@stop.mode = AND' on the 1st line of the STOPFILE


OR mode declared using '@stop.mode = OR' on the 1st line of the STOPFILE.

In AND mode, bigrams whose both constituent words are stop words are removed while, in OR mode, bigrams whose either or both constituent words are stopwords are removed from the output.

--window W

Tokens appearing within W positions from each other (with at most W-2 intervening words) will form bigrams. Same as's --window option.

--remove L

Bigrams with counts less than L in the entire SOURCE data are removed from the sample. The counts of the removed bigrams are not counted in any marginal totals. This has same effect as's --remove option.

--frequency F

Bigrams with counts less than F in the entire SOURCE are not displayed. The counts of the skipped bigrams ARE counted in the marginal totals. In other words, --frequency in has same effect as the's --frequency option.


Switches ON the --newLine option in This will prevent bigrams from spanning across the lines.

Other Options :


Displays this message.


Displays the version information.



Intermediate count results created in STEP 2 are recombined in a pair-wise fashion such that for P separate count output files, C1, C2, C3 ... , CP,

C1 and C2 are first recombined and result is written to huge-count.output

Counts from each of the C3, C4, ... CP are then combined (added) to huge-count.output and each time while recombining, always the smaller of the two files is loaded.


After all files are recombined, the resultant huge-count.output is then sorted in the descending order of the bigram counts. If --remove is specified, bigrams with counts less than the specified value of --remove, in the final huge-count.output file are removed from the sample and their counts are deleted from the marginal totals. If --frequency is selected, bigrams with scores less than the specified value are simply skipped from output.


After huge-count finishes successfully, DESTINATION will contain -

BUGS ^ doesn't consider bigrams at file boundaries. In other words, the result of and on the same data file will differ if --newLine is not used, in that, runs on multiple files separately and thus looses the track of the bigrams on file boundaries. With --window not specified, there will be loss of one bigram at each file boundary while its W bigrams with --window W.

Functionality of huge-count is same as count only if --newLine is used and all files start and end on sentence boundaries. In other words, there should not be any sentence breaks at the start or end of any file given to huge-count.


Amruta Purandare, Ted Pedersen. University of Minnesota at Duluth.


Copyright (c) 2004,

Amruta Purandare, University of Minnesota, Duluth.

Ted Pedersen, University of Minnesota, Duluth.

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program; if not, write to

The Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.

syntax highlighting: