Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,48 @@
|
|
1 |
---
|
2 |
license: gpl
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: gpl
|
3 |
+
language:
|
4 |
+
- hu
|
5 |
+
tags:
|
6 |
+
- text-classification
|
7 |
+
metrics:
|
8 |
+
- accuracy
|
9 |
+
widget:
|
10 |
+
- text: "Kovácsné Nagy Erzsébet [SEP] A Kovácsné Nagy Erzsébet nagyon jól érzi magát a Nokiánál, azonban a Németországból érkezett Kovács Péter nehezen boldogul a beilleszkedéssel."
|
11 |
+
|
12 |
---
|
13 |
+
|
14 |
+
# Hungarian Aspect-based Sentiment Analysis model with XLM-RoBERTa
|
15 |
+
|
16 |
+
For further models, scripts and details, see [our repository](https://github.com/nytud/sentiment-analysis) or [our demo site](https://juniper.nytud.hu/demo/nlp).
|
17 |
+
|
18 |
+
- Pretrained model used: XLM-RoBERTa
|
19 |
+
- Finetuned on OpinHuBank (OHB) Corpus
|
20 |
+
- Labels: 1, 2, 3
|
21 |
+
|
22 |
+
## Limitations
|
23 |
+
|
24 |
+
- max_seq_length = 256
|
25 |
+
|
26 |
+
## Results
|
27 |
+
|
28 |
+
| Model | OHB |
|
29 |
+
| ------------- | ------------- |
|
30 |
+
| huBERT | 82.30 |
|
31 |
+
| XLM-R | 80.59 |
|
32 |
+
|
33 |
+
|
34 |
+
## Citation
|
35 |
+
If you use this model, please cite the following paper:
|
36 |
+
|
37 |
+
```
|
38 |
+
@inproceedings {yang-asent,
|
39 |
+
title = {Neurális entitásorientált szentimentelemző alkalmazás magyar nyelvre},
|
40 |
+
booktitle = {XIX. Magyar Számítógépes Nyelvészeti Konferencia (MSZNY 2023)},
|
41 |
+
year = {2023},
|
42 |
+
publisher = {Szegedi Tudományegyetem},
|
43 |
+
address = {Szeged, Hungary},
|
44 |
+
author = {Yang, Zijian Győző, Laki László János},
|
45 |
+
pages = {0}
|
46 |
+
}
|
47 |
+
|
48 |
+
```
|