# polemo2-official

**Repository Path**: hf-datasets/polemo2-official

## Basic Information

- **Project Name**: polemo2-official
- **Description**: Mirror of https://huggingface.co/datasets/clarin-pl/polemo2-official
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2023-10-18
- **Last Updated**: 2024-06-09

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

---
annotations_creators: 
- expert-generated 
language_creators: 
- other
language: 
- pl
license: 
- cc-by-sa-4.0
multilinguality: 
- monolingual
pretty_name: 'Polemo2'
size_categories: 
- 8K
- 1K<n<10K
source_datasets: 
- original
task_categories: 
- text-classification
task_ids: 
- sentiment-classification
---

# Polemo2

## Description

The PolEmo2.0 is a dataset of online consumer reviews from four domains: medicine, hotels, products, and university. It is human-annotated on a level of full reviews and individual sentences. Current version (PolEmo 2.0) contains 8,216 reviews having 57,466 sentences. Each text and sentence was manually annotated with sentiment in the 2+1 scheme, which gives a total of 197,046 annotations. About 85% of the reviews are from the medicine and hotel domains. Each review is annotated with four labels: positive, negative, neutral, or ambiguous.

## Tasks (input, output and metrics)

The task is to predict the correct label of the review.

**Input** ('*text*' column): sentence

**Output** ('*target*' column): label for sentence sentiment ('zero': neutral, 'minus': negative, 'plus': positive, 'amb': ambiguous)

**Domain**: Online reviews

**Measurements**: Accuracy, F1 Macro

**Example**: 

Input: `Na samym wejściu hotel śmierdzi . W pokojach jest pleśń na ścianach , brudny dywan . W łazience śmierdzi chemią , hotel nie grzeje w pokojach panuje chłód . Wyposażenie pokoju jest stare , kran się rusza , drzwi na balkon nie domykają się . Jedzenie jest w małych ilościach i nie smaczne . Nie polecam nikomu tego hotelu .`

Input (translated by DeepL): `At the very entrance the hotel stinks . In the rooms there is mold on the walls , dirty carpet . The bathroom smells of chemicals , the hotel does not heat in the rooms are cold . The room furnishings are old , the faucet moves , the door to the balcony does not close . The food is in small quantities and not tasty . I would not recommend this hotel to anyone .`

Output:  `1` (negative)

## Data splits

| Subset | Cardinality |
|--------|------------:|
| train  |        6573 |
| val    |         823 |
| test   |         820 |

## Class distribution

| Class   |   train |          dev |   test |
|:--------|--------:|-------------:|-------:|
| minus   |  0.3756 |       0.3694 | 0.4134 |
| plus    |  0.2775 |       0.2868 | 0.2768 |
| amb     |  0.1991 |       0.1883 | 0.1659 |
| zero    |  0.1477 |       0.1555 | 0.1439 |

## Citation

```
@inproceedings{kocon-etal-2019-multi,
    title = "Multi-Level Sentiment Analysis of {P}ol{E}mo 2.0: Extended Corpus of Multi-Domain Consumer Reviews",
    author = "Koco{\'n}, Jan  and
      Mi{\l}kowski, Piotr  and
      Za{\'s}ko-Zieli{\'n}ska, Monika",
    booktitle = "Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL)",
    month = nov,
    year = "2019",
    address = "Hong Kong, China",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/K19-1092",
    doi = "10.18653/v1/K19-1092",
    pages = "980--991",
    abstract = "In this article we present an extended version of PolEmo {--} a corpus of consumer reviews from 4 domains: medicine, hotels, products and school. Current version (PolEmo 2.0) contains 8,216 reviews having 57,466 sentences. Each text and sentence was manually annotated with sentiment in 2+1 scheme, which gives a total of 197,046 annotations. We obtained a high value of Positive Specific Agreement, which is 0.91 for texts and 0.88 for sentences. PolEmo 2.0 is publicly available under a Creative Commons copyright license. We explored recent deep learning approaches for the recognition of sentiment, such as Bi-directional Long Short-Term Memory (BiLSTM) and Bidirectional Encoder Representations from Transformers (BERT).",
}
```

## License

```
Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)
```

## Links

[HuggingFace](https://huggingface.co/datasets/clarin-pl/polemo2-official)

[Source](https://clarin-pl.eu/dspace/handle/11321/710)

[Paper](https://aclanthology.org/K19-1092/)

## Examples

### Loading

```python
from pprint import pprint

from datasets import load_dataset

dataset = load_dataset("clarin-pl/polemo2-official")
pprint(dataset['train'][0])

# {'target': 1,
#  'text': 'Na samym wejściu hotel śmierdzi . W pokojach jest pleśń na ścianach '
#          ', brudny dywan . W łazience śmierdzi chemią , hotel nie grzeje w '
#          'pokojach panuje chłód . Wyposażenie pokoju jest stare , kran się '
#          'rusza , drzwi na balkon nie domykają się . Jedzenie jest w małych '
#          'ilościach i nie smaczne . Nie polecam nikomu tego hotelu .'}
```

### Evaluation

```python
import random
from pprint import pprint

from datasets import load_dataset, load_metric

dataset = load_dataset("clarin-pl/polemo2-official")
references = dataset["test"]["target"]

# generate random predictions
predictions = [random.randrange(max(references) + 1) for _ in range(len(references))]

acc = load_metric("accuracy")
f1 = load_metric("f1")

acc_score = acc.compute(predictions=predictions, references=references)
f1_score = f1.compute(predictions=predictions, references=references, average='macro')

pprint(acc_score)
pprint(f1_score)

# {'accuracy': 0.2475609756097561}
# {'f1': 0.23747048177471738}
```