Dataset Card for "TID-8"
Table of Contents
- Dataset Description
- Dataset Structure
- Dataset Creation
- Considerations for Using the Data
- Additional Information
Dataset Description
- Homepage: placeholder
- Repository: More Information Needed
- Paper: More Information Needed
- Point of Contact: More Information Needed
Dataset Summary
TID-8 is a new benchmark focused on the task of letting models learn from data that has inherent disagreement.
Annotation Split: We split the annotations for each annotator into train and test set.
In other words, the same set of annotators appear in both train, (val), and test sets.
For datasets that have splits originally, we follow the original split and remove datapoints in test sets that are annotated by an annotator who is not in the training set.
For datasets that do not have splits originally, we split the data into train and test set for convenience, you may further split the train set into a train and val set.
Annotator Split: We split annotators into train and test set.
In other words, a different set of annotators would appear in train and test sets.
We split the data into train and test set for convenience, you may consider further splitting the train set into a train and val set for performance validation.
Supported Tasks and Leaderboards
Languages
Dataset Structure
Data Instances
Data Fields
The data fields are the same among all splits. See aforementioned information.
Data Splits
See aforementioned information.
Dataset Creation
Curation Rationale
Source Data
Initial Data Collection and Normalization
Who are the source language producers?
Annotations
Annotation process
Who are the annotators?
Personal and Sensitive Information
Considerations for Using the Data
Social Impact of Dataset
Discussion of Biases
Other Known Limitations
Additional Information
Dataset Curators
Licensing Information
Citation Information
@inproceedings{deng2023tid8,
title={You Are What You Annotate: Towards Better Models through Annotator Representations},
author={Deng, Naihao and Liu, Siyang and Zhang, Frederick Xinliang and Wu, Winston and Wang, Lu and Mihalcea, Rada},
booktitle={Findings of EMNLP 2023},
year={2023}
}
Note that each TID-8 dataset has its own citation. Please see the source to
get the correct citation for each contained dataset.