Code for paper Unsupervised Sentence Textual Similarity with Compositional Phrase Semantics
- python OT
- spacy
- tqdm
- pandas
- numpy
- sklearn
- scipy
- scikits.bootstrap
- prepare data yourself
- prepare the vectors in
vectors
folder, see 'model/word_vector.py' for path specifications and see appendix for downloadable links. - prepare the datasets in
dataset
folder, see 'model/dataset.py' for path specifications and see appendix for ways to obtain and preprocess.
- run pipline.py file for evaluation, note that you may need to set the config files
For sample configs, please see the config
folder.
@inproceedings{wang2022Unsupervised,
title={Unsupervised Sentence Textual Similarity with Compositional Phrase Semantics},
author={Zihao Wang and Jiaheng Dou and Yong Zhang},
booktitle={Proceedings of the 29th International Conference on Computational Linguistics},
year={2022}
}