annotations_creators:
- no-annotation
language_creators:
- found
language:
- ca
license:
- cc-by-nc-4.0
multilinguality:
- monolingual
pretty_name: Racó Forums
task_categories:
- fill-mask
task_ids: []
Dataset Card for Racó Forums Corpus
Table of Contents
- Table of Contents
- Dataset Description
- Dataset Structure
- Dataset Creation
- Considerations for Using the Data
- Additional Information
Dataset Description
- Point of Contact: langtech@bsc.es
Dataset Summary
The Racó Forums Corpus is a 19-million-sentence corpus of Catalan user-generated text built from the forums of Racó Català.
Since the existing available corpora in Catalan lacked conversational data, we searched for a major source of such data for Catalan, and we found Racó Català, a popular multitopic online forum. We obtained a database dump and we transformed all the threads so that we obtained documents that traversed all the existing paths from the root (initial comment) to the leaves (last comment with no reply). In other words, if T is a tree such that T = {A,B,C,D} and the first comment is A that is replied by B and C independently, and, then, C is replied by D, we obtain two different documents A,B and A,C,D in the fairseq language modeling format.
This work is licensed under a Creative Commons Attribution Non-commercial 4.0 International License.
Supported Tasks and Leaderboards
This corpus is mainly intended to pretrain language models and word representations.
Languages
The dataset is in Catalan (ca-ES).
Dataset Structure
The sentences are ordered to preserve the forum structure of comments and answers. T is a tree such that T = {A,B,C,D} and the first comment is A that is replied by B and C independently, and, then, C is replied by D, we obtain two different documents A,B and A,C,D in the fairseq language modeling format.
Data Instances
Ni la Paloma, ni la Razz, ni Bikini, ni res: la cafeteria Slàvia, a Les borges Blanques. Quin concertàs el d'ahir de Pomada!!! Fuà!!! va ser tan tan tan tan tan tan tan bo!!! Flipant!!! Irrepetible!!
És cert, l'Slàvia mola màxim.
Data Splits
The dataset contains two splits: train and valid.
Dataset Creation
Curation Rationale
We created this corpus to contribute to the development of language models in Catalan, a low-resource language. The data was structured to preserve the dialogue structure of forums.
Source Data
Initial Data Collection and Normalization
The data was structured and anonymized by the BSC.
Who are the source language producers?
The data was provided by Racó Català.
Annotations
The dataset is unannotated.
Annotation process
[N/A]
Who are the annotators?
[N/A]
Personal and Sensitive Information
The data was annonymised to remove user names and emails, which were changed to random Catalan names. The mentions to the chat itself have also been changed.
Considerations for Using the Data
Social Impact of Dataset
We hope this corpus contributes to the development of language models in Catalan, a low-resource language.
Discussion of Biases
We are aware that, since the data comes from user-generated forums, this will contain biases, hate speech and toxic content. We have not applied any steps to reduce their impact.
Other Known Limitations
[N/A]
Additional Information
Dataset Curators
Text Mining Unit (TeMU) at the Barcelona Supercomputing Center (bsc-temu@bsc.es).
This work was funded by the Departament de la Vicepresidència i de Polítiques Digitals i Territori de la Generalitat de Catalunya within the framework of Projecte AINA.
Licensing Information
This work is licensed under a Creative Commons Attribution Non-commercial 4.0 International License.
Citation Information
Contributions
Thanks to Racó Català for sharing their data.