irony-detection

URL: https://github.com/zuzannad2/irony-detection

Description: Code used in experiment for CLEF2022’s Shared Task: Profiling Irony and Stereotype Spreaders on Twitter (IROSTEREO)

Methods

• Model:
  • The code focuses on profiling irony in Twitter posts using machine learning techniques.
  • The model aims to classify tweets as ironic or non-ironic, and to identify users spreading stereotypes.
• Approach:
  • The experiment uses a combination of feature extraction from tweet text and machine learning models to classify and profile users.
  • Several classifiers might have been used, potentially Linear SVM, AdaBoost, Random Forest and Ensemble.

Results

• The results are likely part of the CLEF2022 competition, where the focus was on profiling irony and stereotype spreaders.
• The evaluation metrics could include precision, recall, and F1 score, especially for the task of detecting irony in a large volume of social media data.

Dataset

• Dataset Used:
  • The dataset consists of Twitter data, specifically tailored for CLEF2022’s task of profiling irony and stereotype spreaders.
  • Tweet Annotations: Tweets are annotated with irony labels, and some tweets may also contain stereotype-related content.
• Data Processing: Preprocessing may involve cleaning text data, tokenization, and transforming it into a suitable format for training machine learning models.