To clone the repo:
git clone https://github.com/haeunchang/CS51-Final-Project.git
For running the code, Natural Language Toolkit (NLTK) and pyhyphen have to be installed and Python must be able to import both.
NLTK 3.0 can be found here: http://www.nltk.org/
pyhyphen can be found here: https://code.google.com/p/pyhyphen/
For running the code, Python 3 is recommended. Most of the code should be compatible with both Python 2 and Python 3, but because of unicode issues Python 2 is not able to run the code out-of-the-box and additional tweaking may be necessary.
Install all dependencies (including Python 3)
sudo yum install python3 python3-setuptools python3-PyYAML python3-pip python3-devel python3-numpy
Install NLTK: download NLTK 3.0 alpha 3 from http://www.nltk.org/nltk3-alpha/ and extract contents
After extracting cd into directory and run
sudo python3 setup.py install
Add NLTK to PYTHONPATH
PYTHONPATH="/usr/lib/python3.3/site-packages/nltk-3.0a3-py3.3.egg/":"${PYTHONPATH}"
export PYTHONPATH
Install PyHyphen
sudo python3-pip install PyHyphen
Give right (or working) permissions for hyphen directory
sudo chmod -R 755 /usr/lib/python3.3/site-packages/hyphen
Download NLTK data
python3
>>> import nltk
>>> nltk.download()
Downloader> d all
NLTK can be found as community/python-nltk package
sudo pacman -S python-nltk
pyhyphen can be found from AUR: python-pyhyphen package
NLTK also needs to download necessary data, for that
python
>>> import nltk
>>> nltk.download()
Downloader> d all
To install PyHyphen:
Download the zip file found on pyhyphen website.
cd to the temporary directory
Install PyHyphen: sudo -E python3 setup.py install
To install nltk:
Download the zip file found on nltk website for python 3
cd to the temporary directory
Install nltk: sudo python3 setup.py install
To download NLTK data:
In python shell:
>>> import nltk
>>> nltk.download()
Downloader> d all
Usage: main.py [options]
Options:
-h, --help show this help message and exit
-t FILE, --train=FILE
train using data from FILE
-g, --generate generate new haikus using data aquired from training
-f, --fresh overwrite old training databases (old information WILL
be lost)
-m, --markov generate a haiku using a markov chain process
--vocabulary only use training data for training vocabulary
Example usage:
python3 main.py -t data.txt # train using data.txt
python3 main.py -g # generate poems using evolutionary approach
python3 main.py -m # generate poems using markov approach
Here is a description of what each file will do:
- data.txt: cleaned database of haikus (Basho, ...)
- dictionary.py: module for interacting with several external libraries
- training.py: training input data
- evaluate.py: evaluating the goodness of poems
- evolve_population.py: the poem evolution code
- evo_object.py: class Evo_object
- monogram.py: class Monogram
- bigram.py: class Bigram
- line_haiku.py: class Line_Haiku
- line_type.py: class Line_type
- population_haiku.py: class Population_haiku
- main.py: using the program