metadata

language:
  - en
license: cc-by-sa-4.0
size_categories:
  - 1K<n<10K
task_categories:
  - text-classification
  - zero-shot-classification
pretty_name: DocMSU
tags:
  - sarcasm detection
  - sarcasm-detection
  - multi-modal sarcasm detection
dataset_info:
  config_name: sarcasm-detection
  features:
    - name: image
      dtype: image
    - name: text
      dtype: string
    - name: label
      dtype: int64
  splits:
    - name: train
      num_bytes: 1009277842.339
      num_examples: 50279
    - name: validation
      num_bytes: 321454774.496
      num_examples: 14222
    - name: test
      num_bytes: 167725479.432
      num_examples: 7327
  download_size: 1476011526
  dataset_size: 1498458096.267
configs:
  - config_name: sarcasm-detection
    data_files:
      - split: train
        path: sarcasm-detection/train-*
      - split: validation
        path: sarcasm-detection/validation-*
      - split: test
        path: sarcasm-detection/test-*

DocMSU: A Comprehensive Benchmark for Document-level Multimodal Sarcasm Understanding

This is a replication of the DocMSU dataset for easier access.

Reference

Du, H., Nan, G., Zhang, S., Xie, B., Xu, J., Fan, H., Cui, Q., Tao, X. and Jiang, X., 2024, March. DocMSU: A Comprehensive Benchmark for Document-Level Multimodal Sarcasm Understanding. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 38, No. 16, pp. 17933-17941).