For a better user experience, please consider switching to a desktop-friendly view to access this form!
Datasheets Form
Dataset Information
This section requests dataset specific information, including its content and purpose.
Dataset Credentials
This section collects key details on the dataset provider, related research, and access permissions.
Contributor Information
This section gathers details about the dataset contributors.
Available datasheet list can be accessed via https://catalog.hausanlp.org/ Before filling this form, please kindly check if the dataset is present in the list or not.
Name *This will be used for the contributor page.
Email *
Dataset Name *For example: NaijaSenti, AfriSenti, MasakhaNER, etc.
Dataset URL *Direct link to the dataset repository.
Dataset's HuggingFace URL
Link to the HuggingFace version of the dataset (if present). For example: https://huggingface.co/HausaNLP/afrisenti-twi-regression
Dataset Languages *List all the languages in the dataset. For example, NaijaSenti dataset has four languages: Hausa, Yoruba, Igbo, and Pidgin.
1 selected
Dataset Tasks *
1 selected
Dataset Modality *
Dataset Domain *
Dataset Description *
A brief description of the dataset (3-4 long sentences). For example: NaijaSenti dataset is a Twitter-based sentiment analysis resource for four major Nigerian languages: Hausa, Yoruba, Igbo, Nigerian, Pidgin. Each data instance was labeled with one of three sentiment classes: positive, neutral, or negative. The annotation process involved three independent annotators per data instance, with the final sentiment label determined via a majority voting. This dataset provides a robust foundation for multilingual sentiment analysis in underrepresented languages.
Dataset License *
Publish Year *Year of publishing the dataset/paper.
Dataset Collection Style *
Choose
Dataset Quantity/Volume *For example, the NaijaSenti dataset consists of about 30,000 dataset instances, so the field will be 30000.
Dataset Unit *
Choose
Is the dataset split?Whether the dataset is split into train/val/test or train/val.
