Performs dialect analysis using autoencoders. Generates output files that can be ported to GabMap or visualized using SplitsTree.
Taraka Rama and Çağrı Çöltekin. LSTM Autoencoders for Dialect Analysis. Proceedings of Third Workshop on NLP for Similar Languages, Varieties and Dialects (VarDial3), COLING 2016, Osaka, Japan, 2016. http://www.aclweb.org/anthology/W/W16/W16-48.pdf#page=37
- Need Keras (https://keras.io/) with Tensorflow (https://www.tensorflow.org/) as a backend for running the code.
- The main program is in src folder as ae.py
- Run the program as
python3 src/ae.py data/Germany40.utf8
- The program generates a nexus file (.nex extension) and a vectors file in the data folder
In case of any questions, contact [email protected]