File size: 2,225 Bytes
d89bb2d
 
 
 
 
 
 
5d95eb1
ad5fc33
 
 
 
 
 
 
 
d89bb2d
718e4bc
755164c
718e4bc
d89bb2d
 
 
 
 
 
 
844a465
d89bb2d
 
 
 
 
 
 
718e4bc
 
 
72eb418
718e4bc
 
 
d89bb2d
 
 
 
3482915
d89bb2d
718e4bc
 
 
 
 
 
 
 
 
 
 
 
798aa8c
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
---
datasets:
- HiTZ/AbstRCT-ES
language:
- es
- en
pipeline_tag: token-classification
widget:
  - text: >-
      The dysuria resolved faster in patients implanted with 103Pd but was
      unaffected by the use of supplemental radiotherapy and/or androgen
      deprivation therapy.
  - text: >-
      La disuria se resolvió más rápidamente en los pacientes implantados con
      103Pd, pero no se vio afectada por el uso de radioterapia suplementaria
      y/o terapia de privación de andrógenos.
---

# Cross-lingual Argument Mining in the Medical Domain

This model is a fine-tuned version of mBERT for the argument mining task using AbstRCT data in English and Spanish.  
The dataset consists of abstracts of 5 disease types for argument component detection and argument relation classification:

- `neoplasm`: 350 train, 100 dev and 50 test abstracts
- `glaucoma_test`: 100 abstracts
- `mixed_test`: 100 abstracts (20 on glaucoma, 20 on neoplasm, 20 on diabetes, 20 on hypertension, 20 on hepatitis) 

The results (F1 macro averaged at token level) achieved for each test set: 

   Test   | F1-macro | F1-Claim | F1-Premise 
--|-------|-------|-------
 Neoplasm |   82.36  |  74.89   |   89.07    
 Glaucoma |   80.52  |  75.22   |   84.86     
   Mixed  |   81.69  |  75.06   |   88.57  

You can find more information:

 - 📖 Paper: [Crosslingual Argument Mining in the Medical Domain](https://arxiv.org/abs/2301.10527)
 - 💻Code: [https://github.com/ragerri/abstrct-projections](https://github.com/ragerri/abstrct-projections) 

You can load the model as follows:

```python
from transformers import AutoModelForSequenceClassification

model = AutoModelForSequenceClassification.from_pretrained('HiTZ/mbert-argument-mining-es')
````

## Citation

````bibtex
@misc{yeginbergen2024crosslingual,
      title={Cross-lingual Argument Mining in the Medical Domain}, 
      author={Anar Yeginbergen and Rodrigo Agerri},
      year={2024},
      eprint={2301.10527},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}
````

**Contact**: [Anar Yeginbergen](https://ixa.ehu.eus/node/13807?language=en) and [Rodrigo Agerri](https://ragerri.github.io/)
HiTZ Center - Ixa, University of the Basque Country UPV/EHU