File size: 12,995 Bytes
6b255bf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a8b0259
6b255bf
a8b0259
6b255bf
1699c8d
6b255bf
 
 
 
 
 
 
 
 
 
 
 
a8b0259
6b255bf
 
 
 
 
 
a8b0259
 
 
6b255bf
 
 
 
a8b0259
6b255bf
 
 
 
 
 
 
a8b0259
6b255bf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a8b0259
6b255bf
 
a8b0259
 
6b255bf
 
 
a8b0259
 
6b255bf
 
 
a8b0259
 
 
6b255bf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a8b0259
6b255bf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a8b0259
6b255bf
 
 
 
 
 
 
 
 
 
 
 
 
 
a8b0259
 
 
 
 
 
 
 
 
6b255bf
a8b0259
6b255bf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
---
tags:
- setfit
- sentence-transformers
- text-classification
- generated_from_setfit_trainer
widget:
- text: Etunimi Sukunimi Ruotsin kansasta yli puolet reilusti kannattaa Natoon hakemista.
    Ainoana esteenä näkisin, että Ruotsin asevoimat eivät ole läheskään niin hyvällä
    tolalla kuin Suomen mitä tulee varusmiehiin, reserviläisiin tai edes kalustoonkaan.
    Mutta yhteisen hakemuksen kohdalla se tuskin olisi ongelma muille Nato-maille
    hyväksynnän suhteen. Toinen ongelma on, että hyväksynnän tulisi olla sataprosenttinen
    ja ei ole poissujettua, että Venäjä esmes vaikuttaisi yksittäiseen maahan niin,
    että juuri se ei hyväksyisikään hakemusta.
- text: Etunimi Pugh nyt ymmärrän sun puolustelut asut jenkeissä.....
- text: Etunimi Sukunimi ei varmasti moni uskalla
- text: Etunimi Sukunimi Voisitko laittaa tuohon lastentappoväitteeseen mukaan jonkinlaista
    faktaa. Jää muuten melko irralliseksi heitoksi. Ja etkös aiemmin korostanut, että
    maa ei ole sama kuin ihmiset? No mikä on maa tai valtio, se on jäsentensä muodostama.
    Nyt sitten väität, että Ukraina on maana tappanut lapsia 8 vuotta.
- text: Etunimi Sukunimi Historiaa kirjoitetaan vielä maaliskuun 2020 tapahtumista.
metrics:
- metric
pipeline_tag: text-classification
library_name: setfit
inference: true
base_model: TurkuNLP/bert-base-finnish-cased-v1
model-index:
- name: SetFit with TurkuNLP/bert-base-finnish-cased-v1
  results:
  - task:
      type: text-classification
      name: Text Classification
    dataset:
      name: Unknown
      type: unknown
      split: test
    metrics:
    - type: metric
      value: 0.8718253349471188
      name: Metric
---
# Detect Questions in Asynchronous Conversation Comments

## SetFit with TurkuNLP/bert-base-finnish-cased-v1

This is a [SetFit](https://github.com/huggingface/setfit) model that can be used for Text Classification of actions in asynchronous conversation. This particular model detects if a comment includes a question or not. The configuration of the model is that the model is based on only one annotator's annotations (annotator A3). Metric evaluations are based on conservative ground truth (see paper). This SetFit model uses TurkuNLP/bert-base-finnish-cased-v1 as the Sentence Transformer embedding model (using word embeddings). A LogisticRegression instance is used for classification.
The model has been trained using an efficient few-shot learning technique that involves:

1. Fine-tuning a [Sentence Transformer](https://www.sbert.net) with contrastive learning.
2. Training a classification head with features from the fine-tuned Sentence Transformer.

## Model Details

### Model Description
- **Model Type:** SetFit
- **Sentence Transformer body:** [TurkuNLP/bert-base-finnish-cased-v1](https://huggingface.co/TurkuNLP/bert-base-finnish-cased-v1)
- **Classification head:** a [LogisticRegression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html) instance
- **Maximum Sequence Length:** 512 tokens
- **Number of Classes:** 2 classes (contains statement - yes or no)
<!-- - **Training Dataset:** [Unknown](https://huggingface.co/datasets/unknown) -->
<!-- - **Language:** Unknown -->
<!-- - **License:** Unknown -->

### Model Sources

- **Repository:** [GitHub](https://github.com/henniina/Detecting-paired-actions)
- **Paper:** Paakki, H., Toivanen, P. and Kajava K. (2025). Implicit and Indirect: Detecting Face-threatening and Paired Actions in Asynchronous Online Conversations. Northern European Journal of Language Technology (NEJLT), 11(1), pp. 58-83.


### Model Labels
| Label | Examples                                                                                                                                                                                                                                                                                                                                                                                                                                   |
|:------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| 1     | <ul><li>'Etunimi Sukunimi jep. Suomalainen rokottamaton paha, ukrainalainen rokottamaton hyvä. Tää näkyy olevan nyt se mentaliteetti tällä hetkellä...'</li><li>'Etunimi Sukunimi tilastot.'</li><li>'Etunimi Sukunimi myös delta oli suurimmalle osalle myös rokottamattomille lähes oireeton, omikron kuulemma vielä lievempi👏'</li></ul>                                                                                                   |
| 0     | <ul><li>'Etunimi Sukunimi RAutaa rajoille Suomi suureksi ja Viena vapaaksi'</li><li>'Perussuomalaiset siivoamassa keskustelupalstoja, koronakriisiavustuksien avulla? Onhan tämä nyt joku Monty Python -sketsi?'</li><li>'on se hyvä että Kiurussa ei ole miestä vaan Niskavuoren Hetaa joka pistää tuollaisen pojanklopin aisoihin viimeistään silloin kun Vapaavuori on kaltereissa johtaessaan Uuttamaata terveyspaniikkiin.'</li></ul> |

## Evaluation

### Metrics
| Label   | Metric |
|:--------|:-------|
| **5-fold cross-validated F1** | 0.78 |

## Uses

### Direct Use for Inference

First install the SetFit library:

```bash
pip install setfit
```

Then you can load this model and run inference.

```python
from setfit import SetFitModel

# Download from the 🤗 Hub
model = SetFitModel.from_pretrained("Finnish-actions/SetFit-FinBERT1-A3")
# Run inference
preds = model("Etunimi Sukunimi ei varmasti moni uskalla")
```


### Downstream Use

NB. This model has been trained on data coming from Finnish language asynchronous conversations under crisis related news on Facebook. This specific model has been trained to detect whether a comment includes a question or not. It reflects only one of our annotators' label interpretations, so the best use of our models (see our paper) would be to combine a set of models we provide on our Huggingface (Finnish-actions), and use a model ensemble to provide label predictions. It needs to be noted also that the model may not be well applicable outside of its empirical context, so in downstream applications, one should always conduct an evaluation of the model applicability using manually annotated data from that specific context (see our paper for annotation instructions).


### Out-of-Scope Use

Please use this model only for action detection and analysis. Uses of this model and the involved data for generative purposes (e.g. NLG) is prohibited.


## Bias, Risks and Limitations

Note that the model may produce errors. Due to the size of the training dataset, model may not generalize very well even for other novel topics within the same context. Note that model predictions should not be regarded as final judgments e.g. for online moderation purposes, but each case should also be regarded individually if using model predictions to support moderation. Also, the annotations only reflect three (though experienced) annotators' interpretations, so there might be perspectives on data intepretation that have not been taken into account here.
If model is used to support moderation on social media, we recommend that final judgments should always be left for human moderators.


<!--
### Recommendations

*What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
-->

## Training Details

### Training Set Metrics
| Training set | Min | Median  | Max |
|:-------------|:----|:--------|:----|
| Word count   | 1   | 19.6854 | 213 |

| Label | Training Sample Count |
|:------|:----------------------|
| 0     | 263                   |
| 1     | 700                   |

<!--
### Training Hyperparameters
- batch_size: (16, 16)
- num_epochs: (4, 4)
- max_steps: -1
- sampling_strategy: oversampling
- num_iterations: 6
- body_learning_rate: (2e-05, 1e-05)
- head_learning_rate: 0.01
- loss: CosineSimilarityLoss
- distance_metric: cosine_distance
- margin: 0.25
- end_to_end: False
- use_amp: False
- warmup_proportion: 0.1
- l2_weight: 0.01
- seed: 42
- evaluation_strategy: epoch
- eval_max_steps: -1
- load_best_model_at_end: False

### Training Results
| Epoch  | Step | Training Loss | Validation Loss |
|:------:|:----:|:-------------:|:---------------:|
| 0.0014 | 1    | 0.2224        | -               |
| 0.0692 | 50   | 0.2676        | -               |
| 0.1383 | 100  | 0.2486        | -               |
| 0.2075 | 150  | 0.2208        | -               |
| 0.2766 | 200  | 0.1892        | -               |
| 0.3458 | 250  | 0.1509        | -               |
| 0.4149 | 300  | 0.1194        | -               |
| 0.4841 | 350  | 0.0745        | -               |
| 0.5533 | 400  | 0.039         | -               |
| 0.6224 | 450  | 0.0298        | -               |
| 0.6916 | 500  | 0.01          | -               |
| 0.7607 | 550  | 0.006         | -               |
| 0.8299 | 600  | 0.0021        | -               |
| 0.8990 | 650  | 0.0017        | -               |
| 0.9682 | 700  | 0.0038        | -               |
| 1.0    | 723  | -             | 0.2008          |
| 1.0373 | 750  | 0.0088        | -               |
| 1.1065 | 800  | 0.0041        | -               |
| 1.1757 | 850  | 0.0067        | -               |
| 1.2448 | 900  | 0.0041        | -               |
| 1.3140 | 950  | 0.0021        | -               |
| 1.3831 | 1000 | 0.0036        | -               |
| 1.4523 | 1050 | 0.0036        | -               |
| 1.5214 | 1100 | 0.0011        | -               |
| 1.5906 | 1150 | 0.0035        | -               |
| 1.6598 | 1200 | 0.0047        | -               |
| 1.7289 | 1250 | 0.0005        | -               |
| 1.7981 | 1300 | 0.0002        | -               |
| 1.8672 | 1350 | 0.0029        | -               |
| 1.9364 | 1400 | 0.0029        | -               |
| 2.0    | 1446 | -             | 0.2342          |
| 2.0055 | 1450 | 0.0014        | -               |
| 2.0747 | 1500 | 0.0023        | -               |
| 2.1438 | 1550 | 0.0022        | -               |
| 2.2130 | 1600 | 0.0014        | -               |
| 2.2822 | 1650 | 0.0024        | -               |
| 2.3513 | 1700 | 0.0035        | -               |
| 2.4205 | 1750 | 0.0014        | -               |
| 2.4896 | 1800 | 0.0022        | -               |
| 2.5588 | 1850 | 0.0025        | -               |
| 2.6279 | 1900 | 0.0003        | -               |
| 2.6971 | 1950 | 0.0042        | -               |
| 2.7663 | 2000 | 0.0014        | -               |
| 2.8354 | 2050 | 0.0003        | -               |
| 2.9046 | 2100 | 0.0022        | -               |
| 2.9737 | 2150 | 0.0031        | -               |
| 3.0    | 2169 | -             | 0.2224          |
| 3.0429 | 2200 | 0.0016        | -               |
| 3.1120 | 2250 | 0.0014        | -               |
| 3.1812 | 2300 | 0.005         | -               |
| 3.2503 | 2350 | 0.0045        | -               |
| 3.3195 | 2400 | 0.001         | -               |
| 3.3887 | 2450 | 0.0012        | -               |
| 3.4578 | 2500 | 0.0004        | -               |
| 3.5270 | 2550 | 0.0013        | -               |
| 3.5961 | 2600 | 0.0022        | -               |
| 3.6653 | 2650 | 0.0009        | -               |
| 3.7344 | 2700 | 0.0018        | -               |
| 3.8036 | 2750 | 0.0015        | -               |
| 3.8728 | 2800 | 0.0019        | -               |
| 3.9419 | 2850 | 0.0025        | -               |
| 4.0    | 2892 | -             | 0.2222          |
-->

### Framework Versions
- Python: 3.11.9
- SetFit: 1.1.3
- Sentence Transformers: 3.2.0
- Transformers: 4.44.0
- PyTorch: 2.4.0+cu124
- Datasets: 2.21.0
- Tokenizers: 0.19.1

## Citation

### BibTeX
```bibtex
@article{paakki-implicit-indirect,
    doi = {https://doi.org/10.3384/nejlt.2000-1533.2025.5980},
    url = {https://nejlt.ep.liu.se/article/view/5980},
    author = {Paakki, Henna and Toivanen, Pihla and Kajava, Kaisla},
    title = {Implicit and Indirect: Detecting Face-threatening and Paired Actions in Asynchronous Online Conversations},
    publisher = {Northern European Journal of Language Technology (NEJLT)},
    volume= {11},
    number= {1},
    year = {2025}
}

```

<!--
## Glossary

*Clearly define terms in order to be accessible across audiences.*
-->

<!--
## Model Card Authors

*Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
-->

<!--
## Model Card Contact

*Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
-->