patrickjamesmarcellana/synthetic-filipino-sarcasm-detection

这是一个包含菲律宾语讽刺和非讽刺推文的合成和有限现实世界数据集。数据集由两种语言模型生成的讽刺和非讽刺推文组成，分别是GPT-4o和Gemini 2.0 Flash。合成数据包含504条讽刺推文和504条非讽刺推文。现实世界数据是从2016年菲律宾国家选举期间的推文中提取的，每个推文都经过手动重新标注以识别是否存在讽刺。数据集分为两个配置：合成数据和现实世界数据。

This is a dataset composed of Filipino sarcastic and non-sarcastic tweets, divided into synthetic data generated by language models and limited real-world data extracted from tweets. The synthetic data consists of 504 sarcastic tweets and 504 non-sarcastic tweets generated by GPT-4o and Gemini 2.0 Flash. The real-world data is extracted from tweets during the 2016 Philippine National Elections and each tweet is manually relabeled to identify the presence of sarcasm. The dataset is available in two configurations: synthetic_data and real_world_data.