QWSD

Code and data for the paper "A Quantum-like approach to Word Sense Disambiguation" at RANLP2019.

Tested using:

• Python 3.6.3
• Numpy 1.15.4
• Scipy 0.19.1
• NLTK 3.2.5
• lxml 4.3.0

For reproducing our results:

• Clone the repository.
• Install the NLTK WordNet corpus package.
• Download complex embeddings from this link.
• Classify all the instances taken from the standard benchmark created by [Raganato et al. 2017].

    python3 QWSD.py TestX9c.bin Evaluation_Datasets/ALL/ALL.data.xml > Test

• Evaluate the results:

    bash ./ScoreAll.sh Test

If you find this code useful in your research, please cite:

@inproceedings{Tamburini:2019:RANLP,
  author    = {Fabio Tamburini},
  editor    = {Ruslan Mitkov and
               Galia Angelova},
  title     = {A Quantum-Like Approach to Word Sense Disambiguation},
  booktitle = {Proceedings of the International Conference on Recent Advances in
               Natural Language Processing, {RANLP} 2019, Varna, Bulgaria, September
               2-4, 2019},
  pages     = {1176--1185},
  publisher = {{INCOMA} Ltd.},
  year      = {2019},
  url       = {https://doi.org/10.26615/978-954-452-056-4\_135},
  doi       = {10.26615/978-954-452-056-4\_135},
}

In case of problems contact me at [email protected].