Github repository for my ACL-SRW paper. My paper looks into the problem of finding a cognate from a search space of a lexicon list. I approached the solution by creating a ranker function consisting of two modules: shingle similarity function and graphical error modelling function.

You need Python 3 and Numpy for this.

Currently this takes cognates from four languages:

• Spanish
• Portuguese
• French
• Italian

You can find training, testing and lexicons in the data folder.

You will need Python 3 to run the notebooks.

1. This notebook refers to the shingling concepts which refers to section 2 of the paper.
2. This notebook refers to the construction of graphical error model, which refers to section 3 of the paper.
3. This notebook refers to the string similarity concepts, which refers to section 4.1 of the paper.
4. This notebook refers to the string dis-similarity concepts, which refers to section 4.2 of the paper.
5. This notebook refers to the scoring concepts which refers to section 4.3 of the paper.

Simply run python3 demo.py to demonstrate results from portuguese cognates.

If you find this useful, then please cite my work:

@InProceedings{P18-3019,
  author = 	"A, Pranav",
  title = 	"Alignment Analysis of Sequential Segmentation of Lexicons to Improve Automatic Cognate Detection",
  booktitle = 	"Proceedings of ACL 2018, Student Research Workshop",
  year = 	"2018",
  publisher = 	"Association for Computational Linguistics",
  pages = 	"134--140",
  location = 	"Melbourne, Australia",
  url = 	"http://aclweb.org/anthology/P18-3019"
}