data-of-multimodal-sarcasm-detection

URL: https://github.com/headacheboy/data-of-multimodal-sarcasm-detection

Description: No description provided.

Project Overview

This repository contains text and image data for multimodal sarcasm detection.

Training Methods:

• Text Data:
  • In the training set, the last element in each list represents the label.
  • In the validation and testing sets, two labels are included:
    • The last but one element is labeled based on the original post hashtag.
    • The last element is labeled by a human annotator.
• Preprocessing:
  • Some posts are excluded if they contain specific words such as:
    • exgag, sarcasm, sarcastic, reposting, joke, humour, humor, jokes, irony, ironic.
  • The corresponding images of these posts are also not uploaded.

Results:

• No explicit model performance metrics or evaluation results are provided in the repository.

Dataset:

• Contains both text and image data, but some images are missing due to filtering.