Text related annotation
Human post-editing of Arabic speech transcriptions by listening to each audio segment and correcting the text to match what is spoken.
Human annotation of Arabic spoken utterances with dialogue act and emotional tone, based strictly on the surface linguistic content.
Human verification of Arabic numerical factual claims by judging fluency and faithfulness on a 5-point Likert scale, based on the provided evidence and model justification.
Subjectivity Annotation of the Arabic news articles
Human annotation of Bangla (and Bangla/English-mixed) posts for polarization.
Human annotation of Bangla (and Bangla/English-mixed) posts for polarization.
Judging the explanation of the hateful meme
Verify 3 question–answer pairs for each image. For each pair: (1) mark whether the Q/A is correct and relevant to the image, and (2) mark whether the question cannot be answered without seeing the image.
Verify 3 question–answer pairs for each image. For each pair: (1) mark whether the Q/A is correct and relevant to the image, and (2) mark whether the question cannot be answered without seeing the image.
Edit Egyptian dialect translations of MSA MCQs. MSA fields are fixed and non-editable; only Egyptian fields can be updated.
Edit Levantine Arabic translations of MSA MCQs. MSA fields are fixed and non-editable; only Levantine Arabic fields can be updated.
Record 3 hallucination-related statements for each multimodal task item.