Creating and Evaluating Figurative Language Dataset for Sindhi

1. 1Collect raw text from various sources such as blogs and social media platforms
2. 2Prepare the corpus for annotation using tools like Doccano
3. 3Label the data using native annotators to achieve high inter-annotator agreement
4. 4Establish baseline results using cross-validation techniques like 5-fold and 10-fold cross-validation
5. 5Evaluate the performance of the dataset using metrics like accuracy and F1-score

NLP researchers and developers working with low-resource languages like Sindhi can benefit from this dataset to improve their models' performance

💡 Creating a high-quality dataset for figurative language classification is crucial for improving NLP models' performance in low-resource languages like Sindhi

📊 Introducing SiNFluD, a novel benchmark dataset for Sindhi figurative language classification! 🚀