CADE

CADE is a tool to align distributional vector spaces. CADE is the more general name for our framework Temporal Word Embeddings with a Compass (TWEC). TWEC was originally implemented by Valerio Di Carlo during his masters’ thesis and published at the AAAI2019 Conference.

Eventually, we changed the name due to the more general nature of the alignment provided by the model.

CADE can and has been used for several different tasks: from general temporal vector space alignment [1] and a more general comparison of language variation [2], to tasks like semantic change detection in diachronic contexts [3,6] and narrative understanding [5].

Compass Aligned Distributional Embeddings

You can find more about CADE on the github page and on this blog post.

Our Papers on CADE

[1] Di Carlo, V., Bianchi, F., & Palmonari, M. (2019). Training Temporal Word Embeddings with a Compass. Proceedings of the AAAI Conference on Artificial Intelligence, 33(01), 6326-6334. https://doi.org/10.1609/aaai.v33i01.33016326
[2] Bianchi, F., Di Carlo, V., Nicoli, P., & Palmonari, M. (2020). Compass-aligned Distributional Embeddings for Studying Semantic Differences across Corpora. Arxiv. https://arxiv.org/abs/2004.06519
[3] Belotti, F., Bianchi, F., & Palmonari, M. (2020). UNIMIB @ DIACR-Ita: Aligning Distributional Embeddings with a Compass for Semantic Change Detection in the Italian Language. Proceedings of Clic-it (to appear).
[4] Cassani, G., Bianchi, F., Marelli, M., (2020). Words that shift are learned later: A study on the relation between diachronic semantic change and age of acquisition. AMLap.; video

Other works using CADE

[5] Temporal Narrative Understanding Volpetti, C., Vani, K., & Antonucci, A. (2020, February). Temporal Word Embeddings for Narrative Understanding. In Proceedings of the 2020 12th International Conference on Machine Learning and Computing (pp. 68-72).
[6] Showing CADE effectiveness for semantic change detection Alkhalifa R., and Tsakalidis A., Zubiaga A., & Liakata, M. (2020). QMUL-SDS @ DIACR-ITA2020: Evaluating Unsupervised Diachronic Lexical Semantics Classification in Italian. Proceedings of Clic-it (to appear).

Main References

If you use CADE in a research paper, please cite the two following works:

@inproceedings{di2019training,
  title={Training temporal word embeddings with a compass},
  author={Di Carlo, Valerio and Bianchi, Federico and Palmonari, Matteo},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={33},
  pages={6326--6334},
  year={2019}
}

@article{bianchi2020compass,
  title={Compass-aligned Distributional Embeddings for Studying Semantic Differences across Corpora},
  author={Bianchi, Federico and Di Carlo, Valerio and Nicoli, Paolo and Palmonari, Matteo},
  journal={arXiv preprint arXiv:2004.06519},
  year={2020}
}