add extra paper information
Browse files
README.md
CHANGED
@@ -41,10 +41,28 @@ dataset_info:
|
|
41 |
license: apache-2.0
|
42 |
---
|
43 |
|
|
|
|
|
|
|
|
|
|
|
44 |
# PWESuite-Eval
|
45 |
|
46 |
Dataset composed of multiple smaller datasets used for the evaluation of phonetic word embeddings.
|
47 |
See code for evaluation [here](https://github.com/zouharvi/pwesuite).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
|
49 |
Used datasets:
|
50 |
- [CMU Pronunciation dictionary](http://www.speech.cs.cmu.edu/cgi-bin/cmudict)
|
@@ -58,16 +76,4 @@ Authors:
|
|
58 |
- Chenxuan Cui (CMU LTI, [contact](mailto:[email protected]))
|
59 |
- Nathaniel Robinson (CMU LTI, [contact](mailto:[email protected]))
|
60 |
- Nathaniel Carlson (BYU, [contact](mailto:[email protected]))
|
61 |
-
- David Mortensen (CMU LTI, [contact](mailto:[email protected]))
|
62 |
-
|
63 |
-
If you use this dataset/evaluation, please cite:
|
64 |
-
|
65 |
-
```
|
66 |
-
@article{zouhar2023pwesuite,
|
67 |
-
title={{PWESuite}: {P}honetic Word Embeddings and Tasks They Facilitate},
|
68 |
-
author={Zouhar, Vil{\'e}m and Chang, Kalvin and Cui, Chenxuan and Carlson, Nathaniel and Robinson, Nathaniel and Sachan, Mrinmaya and Mortensen, David},
|
69 |
-
journal={arXiv preprint arXiv:2304.02541},
|
70 |
-
year={2023},
|
71 |
-
url={https://arxiv.org/abs/2304.02541}
|
72 |
-
}
|
73 |
-
```
|
|
|
41 |
license: apache-2.0
|
42 |
---
|
43 |
|
44 |
+
|
45 |
+
<p align="center">
|
46 |
+
<img src="https://github.com/zouharvi/pwesuite/assets/7661193/919ab8b2-f635-4a24-8a23-9b8ab8663e8d" width="500em">
|
47 |
+
</p>
|
48 |
+
|
49 |
# PWESuite-Eval
|
50 |
|
51 |
Dataset composed of multiple smaller datasets used for the evaluation of phonetic word embeddings.
|
52 |
See code for evaluation [here](https://github.com/zouharvi/pwesuite).
|
53 |
+
If you use this dataset/evaluation, please cite the [paper at LREC-COLING 2024](https://arxiv.org/abs/2304.02541):
|
54 |
+
|
55 |
+
```
|
56 |
+
@article{zouhar2023pwesuite,
|
57 |
+
title={{PWESuite}: {P}honetic Word Embeddings and Tasks They Facilitate},
|
58 |
+
author={Zouhar, Vil{\'e}m and Chang, Kalvin and Cui, Chenxuan and Carlson, Nathaniel and Robinson, Nathaniel and Sachan, Mrinmaya and Mortensen, David},
|
59 |
+
journal={arXiv preprint arXiv:2304.02541},
|
60 |
+
year={2023},
|
61 |
+
url={https://arxiv.org/abs/2304.02541}
|
62 |
+
}
|
63 |
+
```
|
64 |
+
|
65 |
+
> **Abstract:** Mapping words into a fixed-dimensional vector space is the backbone of modern NLP. While most word embedding methods successfully encode semantic information, they overlook phonetic information that is crucial for many tasks. We develop three methods that use articulatory features to build phonetically informed word embeddings. To address the inconsistent evaluation of existing phonetic word embedding methods, we also contribute a task suite to fairly evaluate past, current, and future methods. We evaluate both (1) intrinsic aspects of phonetic word embeddings, such as word retrieval and correlation with sound similarity, and (2) extrinsic performance on tasks such as rhyme and cognate detection and sound analogies. We hope our task suite will promote reproducibility and inspire future phonetic embedding research.
|
66 |
|
67 |
Used datasets:
|
68 |
- [CMU Pronunciation dictionary](http://www.speech.cs.cmu.edu/cgi-bin/cmudict)
|
|
|
76 |
- Chenxuan Cui (CMU LTI, [contact](mailto:[email protected]))
|
77 |
- Nathaniel Robinson (CMU LTI, [contact](mailto:[email protected]))
|
78 |
- Nathaniel Carlson (BYU, [contact](mailto:[email protected]))
|
79 |
+
- David Mortensen (CMU LTI, [contact](mailto:[email protected]))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|