MHSRC

URL: https://github.com/David-deng-01/MHSRC

Description: Multimodal Humor and Sarcasm Recognition in Chinese: Dataset Construction, Association Analysis and Fine-grained Analysis

Dataset

• Name: MHSRC (Multimodal Humor and Sarcasm Recognition in Chinese)
• Size: The dataset details are not specified, but files are provided for download and further processing.

Download Links

• BaiduNetDisk: Links for dataset files including multimodal features and JSON files are provided through Baidu NetDisk (specific download links not listed in the description).

Dataset Details

The dataset is specifically designed for multimodal humor and sarcasm recognition in Chinese, featuring different types of data processing tasks:

1. Dataset Splitting:
   • Download the pre-split dataset [BaiduNetDisk].
   • Alternatively, users can split the data using the script 1_split_data.py.
2. Cache Generation:
   • Cache files are provided for training and validation, which can be downloaded from [BaiduNetDisk].
   • Alternatively, generate cache files by running the script 2_generate_cache.py, considering oversampling for training and undersampling for validation and test sets.
3. Class Cache Generation:
   • For category classification tasks, download processed cache files or generate them using 3_generate_classifier_cache.py.
4. Small Model Training:
   • Users can train small models by running the script 4_train_small_model.py, after ensuring that cache files are generated in the previous steps.
5. Multitask Model Training:
   • Multitask models can be trained with the script 5_train_small_model.py, utilizing pre-processed data.
6. Category Classifier Model Training:
   • To train category classifiers, run 6_train_type_classifier_model.py after preparing the data.

Additional Information

• Objective: The dataset is used for recognizing humor and sarcasm in Chinese text in a multimodal context. It includes tasks like feature extraction, dataset splitting, and training various models for classification.

• Tasks: The dataset supports fine-grained analysis of humor and sarcasm, including category classification and multimodal analysis.