- This is a snapshot of program sources for IJCAI 16, an top international AI conference.
- See our paper.
- This has not been refactored and will not be maintained.
author: Tatsuya Iwanari
Move to src folder and run a command make
.
In src folder, type the following command
./main mode input_file output_path [options]
mode is an integer parameter to specify what the program will do.
value | mode name | explanation |
---|---|---|
0 | count/reduce/format | do all processes. |
1 | reduce/format | do from the second process. |
2 | format | do the last process only. |
input_file differs based on the mode parameter.
mode value | input_file |
---|---|
0 | a list file of social data files. |
1 | a list file of counted files. |
2 | a file to be formatted in svm rank format. |
A list file is composed of lines, each line expresses a file like this:
/path/to/data0
/path/to/data1
...
output_path is a string parameter to specify the output folder. All the output of this program will be output into it.
process | location |
---|---|
count | output_path/count |
reduce | output_path/reduce |
format | output_path/format |
options are mode dependent options.
mode value | options |
---|---|
0 | adjective concept_file [pattern_file_path] |
1 | [none] |
2 | [none] |
adjective is a string parameter which specifies adjective. concept_file is the same format as input_file. pattern_file_path is a path to pattern files for its counter.
I checked it will run on Linux (CentOS release 5.11 (Final)).
This program has a depency on liblzma to decode .xz data files and it needs the following programs:
- Autoconf 2.64
- Automake 1.12
- gettext 0.18 (Note: autopoint depends on cvs!)
- libtool 2.2
If you do not use .xz files, you can remove xz submodule.