README.md

---
license:
- cc-by-4.0
multilinguality:
- monolingual
- aligned
task_categories:
- text-classification
- text2text-generation
source_datasets:
- original
- >-
  extended|other-turkcorpus,other-asset,other-questeval,other-simplicity_da,other-simp_da
language:
- en
tags:
- simplification-evaluation
- meaning-evaluation
pretty_name: CSMD
size_categories:
- 1K<n<10K
dataset_info:
- config_name: meaning
  features:
  - name: original
    dtype: string
  - name: simplification
    dtype: string
  - name: label
    dtype: float64
  splits:
  - name: train
    num_bytes: 251558
    num_examples: 853
  - name: dev
    num_bytes: 27794
    num_examples: 95
  - name: test
    num_bytes: 117686
    num_examples: 407
  download_size: 397038
  dataset_size: 1355
- config_name: meaning_with_data_augmentation
  features:
  - name: original
    dtype: string
  - name: simplification
    dtype: string
  - name: label
    dtype: float64
  splits:
  - name: train
    num_bytes: 1242540
    num_examples: 4267
  - name: dev
    num_bytes: 134726
    num_examples: 475
  - name: test
    num_bytes: 592052
    num_examples: 2033
  download_size: 1969318
  dataset_size: 6775
- config_name: meaning_holdout_identical
  features:
  - name: original
    dtype: string
  - name: simplification
    dtype: string
  - name: label
    dtype: float64
  splits:
  - name: test
    num_bytes: 89866
    num_examples: 359
  download_size: 89866
  dataset_size: 359
- config_name: meaning_holdout_unrelated
  features:
  - name: original
    dtype: string
  - name: simplification
    dtype: string
  - name: label
    dtype: float64
  splits:
  - name: test
    num_bytes: 247835
    num_examples: 359
  download_size: 247835
  dataset_size: 359
config_names:
- meaning
- meaning_with_data_augmentation
- meaning_holdout_identical
- meaning_holdout_unrelated
viewer: true

configs:
- config_name: meaning
  data_files:
  - split: train
    path: "train.tsv"
  - split: dev
    path: "dev.tsv"
  - split: test
    path: "test.tsv"
- config_name: meaning_with_data_augmentation
  data_files:
  - split: train
    path: "train_da.tsv"
  - split: dev
    path: "dev_da.tsv"
  - split: test
    path: "test_da.tsv"
- config_name: meaning_holdout_identical
  data_files:
  - split: test
    path: "identical.tsv"
- config_name: meaning_holdout_unrelated
  data_files:
  - split: test
    path: "unrelated.tsv"
---

# Dataset Card for "Continuous Scale Meaning Dataset" (CSMD)

CSMD was created for [MeaningBERT: Assessing Meaning Preservation Between Sentences](https://www.frontiersin.org/articles/10.3389/frai.2023.1223924/full).

It contains 1,355 English text simplification meaning preservation annotations. Meaning preservation measures how well the meaning of the output text corresponds to the meaning of the source ([Saggion, 2017](https://link.springer.com/book/10.1007/978-3-031-02166-4)).

The annotations were taken from the following four datasets: 

- [ASSET](https://aclanthology.org/2020.acl-main.424/)
- [QuestEVal](https://arxiv.org/abs/2104.07560),
- [SimpDa_2022](https://aclanthology.org/2023.acl-long.905.pdf) and,
- [Simplicity-DA](https://direct.mit.edu/coli/article/47/4/861/106930/The-Un-Suitability-of-Automatic-Evaluation-Metrics).

It contains a data augmentation subset of 1,355 identical sentence triplets and 1,355 unrelated sentence triplets (See the "Sanity Checks" section (3.3.) in our [article](https://www.frontiersin.org/articles/10.3389/frai.2023.1223924/full)).

It also contains two holdout subsets of 359 identical sentence triplets and 359 unrelated sentence triples (See the "MeaningBERT" section (3.4.) in our [article](https://www.frontiersin.org/articles/10.3389/frai.2023.1223924/full)).


## Dataset Structure

### Data Instances

- `Meaning` configuration: an instance consists of 1,355 meaning preservation triplets (Document, simplification, label).
- `meaning_with_data_augmentation` configuration: an instance consists of 1,355 meaning preservation triplets (Document, simplification, label) along with 1,355 data augmentation triplets (Document, Document, 100) and 1,355 data augmentation triplets (Document, Unrelated Document, 0) (See the sanity checks in our [article](https://www.frontiersin.org/articles/10.3389/frai.2023.1223924/full)).
- `meaning_holdout_identical` configuration: an instance consists of 359 meaning holdout preservation identical triplets (Document, Document, 1) based on the ASSET Simplification dataset.
- `meaning_holdout_unrelated` configuration: an instance consists of 359 meaning holdout preservation unrelated triplets (Document, Unrelated Document, 0) based on the ASSET Simplification dataset.

### About the Data Augmentation
#### Unrelated Sentence
We have changed the data augmentation approach for the unrelated sentence. Instead of generating noisy sentences using an LLM, for each of the 1,355 sentences, we sample a sentence in the unlabeled sentence in ASSET (non included in the holdout nor the labelled sentence). We compute the  Rouge1, Rouge2, RougeL and bleu scores to validate that the sentences are unrelated in terms of vocabulary. Namely, each metric score is below 0.20 or 20 for Bleu for all pairs. If a pair achieves a higher value, we select another sentence from ASSET to create a pair and reapply the test until a pair achieves a score below 0.20/20.
### Commutative Property
Since meaning preservation is a commutative function, i.e., Meaning(Sent_a, Sent_b) = Meaning(Sent_b, Sent_a), we also include the commutative version of the triplet in the data augmentation version of the dataset for sentences that are not identical.

### Data Fields

- `original`: an original sentence from the source datasets.
- `simplification`:  a simplification of the original obtained by an automated system or a human.
- `label`: a meaning preservation rating between 0 and 100.

### Data Splits
The split statistics of CSMD are given below.

| | Train    | Dev    | Test | Total |
| ------ | ------   | ------ | ---- | ----- |
| Meaning | 853    | 95   | 407  | 1,355  |
| Meaning With Data Augmentation | 2,560    | 285   | 1,220  | 4,065  |
| Meaning Holdout Identical | NA    | NA   | 359  | 359 |
| Meaning Holdout Unrelated | NA    | NA   | 359  | 359  |

All the splits are randomly split using a 60-10-30 split with the seed `42`.

# Citation Information

```
@ARTICLE{10.3389/frai.2023.1223924,
AUTHOR={Beauchemin, David and Saggion, Horacio and Khoury, Richard},   
TITLE={{MeaningBERT: Assessing Meaning Preservation Between Sentences}},      
JOURNAL={Frontiers in Artificial Intelligence},      
VOLUME={6},           
YEAR={2023},      
URL={https://www.frontiersin.org/articles/10.3389/frai.2023.1223924},       
DOI={10.3389/frai.2023.1223924},        
ISSN={2624-8212},   
}
```