To use this project, following packages need to be installed:
- scikit-learn (using version 0.21.0)
- joblib (0.13.2)
- Theano (1.0.4) or tensorflow (1.10.0)
- Keras (2.2.4)
- numpy (1.16.3)
Optional (for drawing plots):
- matplotlib (used version 3.0.3)
If recollecting data from sources and rebuilding datasets following libraries are needed:
- re
- csv
- xml
- random
Using different version for joblib can render models unusable.
Package data contains source code used for extracting data from sources. Example usage is shown in script classifiers_main.py. One of datasets is in folder sources, others will be on Dropbox (link will be added in a day or two) because files were too big to upload on Github.
Package classifiers contains trained models in folder models and training and testing sets in folder datasets. It also contains source code used when searching for best parameters and saving trained models. Example usage is shown in script classifiers_main.py.
Package evaluation contains source code for evaluation of models. Originally used code is evaluate.py, corrected code is evaluate_corrected.py. Example usage is shown in script evaluation_main.py.
To use modules, one can import them in terminal and use immediately or write to one of scripts that end with main.py and then execute script.
Extracting data again from sources is possible, but I do not provide source files.
Synonym sources:
Antonym sources:
Jasmina Pegan
To delo je objavljeno pod licenco Creative Commons Priznanje avtorstva-Deljenje pod enakimi pogoji 4.0 Mednarodna.
