Datasets:

Skhaled
/

Ar-MUSA

task_categories:
  - text-classification
  - audio-classification
  - image-classification
language:
  - ar
size_categories:
  - 1K<n<10K
license: afl-3.0
tags:
  - multimodal
  - sentiment
  - analysis
  - arabic
  - Ar-MUSA

Data Directory Structure

The Ar-MUSA directory contains annotated datasets organized by batches and annotation teams. Each batch is labeled with a number, and the annotation team is indicated by a letter. The structure is as follows:

Ar-MUSA
├── Annotation 1a
│   ├── frames        # Contains the extracted frames for each record
│   ├── audios        # Contains the corresponding audio files
│   ├── transcripts   # Contains the transcripts of the audio files
│   └── annotations.csv  # CSV file with annotations for each record
│
├── Annotation 1b
│   ├── frames        # Contains the extracted frames for each record
│   ├── audios        # Contains the corresponding audio files
│   ├── transcripts   # Contains the transcripts of the audio files
│   └── annotations.csv  # CSV file with annotations for each record
│
└── Annotation 2a
    ├── frames        # Contains the extracted frames for each record
    ├── audios        # Contains the corresponding audio files
    ├── transcripts   # Contains the transcripts of the audio files
    └── annotations.csv  # CSV file with annotations for each record

Explanation:

Annotation Batches (1a, 1b, 2a, etc.):
- The number represents the batch number (e.g., Batch 1, Batch 2).
- The letter indicates the team responsible for the annotation (e.g., Team A, Team B).

Contents of Each Batch:

frames/: A folder containing extracted video frames for each record.
audios/: A folder with the corresponding audio files for the annotated records.
transcripts/: A folder containing the text transcripts of the audio files.
annotations.csv: A CSV file that includes the annotations for each record, detailing sentiment labels, sarcasm markers, and other relevant metadata.

License

The AR-MUSA dataset is licensed under the Academic Free License 3.0 (afl-3.0) and is provided for research purposes only. Any use of this dataset must comply with the terms of this license.

Citation

If you use the AR-MUSA dataset in your research, please cite the following paper:

@article{khaled2025ar, title={AR-MUSA: a multimodal benchmark dataset and evaluation framework for Arabic sentiment analysis}, author={Khaled, S. and Ragab, M. E. and Helmy, A. K. and Medhat, W. and Mohamed, E. H.}, journal={International Journal of Intelligent Engineering and Systems}, volume={18}, number={4}, pages={30-44}, year={2025}, doi={10.22266/ijies2025.0531.03} }