PortugueseEmotionRecognitionWeakSupervision

本研究创建了一个名为PortugueseEmotionRecognitionWeakSupervision的细粒度情感识别数据集，包含49,179条葡萄牙语推文。数据集通过弱监督方法构建，使用词汇项作为标签规则。数据来源于Twitter，通过API收集，旨在解决低资源环境下情感识别的问题。数据集创建过程中，定义了28种情感类别，并通过人工审核确保情感定义与葡萄牙语的一致性。该数据集适用于自然语言处理领域，特别是情感分析和情感识别任务。

This study develops a fine-grained emotion recognition dataset named PortugueseEmotionRecognitionWeakSupervision, which contains 49,179 Portuguese-language tweets. The dataset is constructed using a weak supervision approach, where lexical items serve as labeling rules. Collected from Twitter via its API, this dataset is intended to tackle the challenge of emotion recognition in low-resource scenarios. During the dataset creation process, 28 emotion categories were defined, and manual reviews were conducted to ensure the consistency between the emotion definitions and the Portuguese language. This dataset is applicable to the field of natural language processing, particularly for sentiment analysis and emotion recognition tasks.