Balanced Arabic Readability Evaluation Corpus (BAREC)
收藏资源简介:
BAREC是一个大规模的细粒度阿拉伯语可读性评估数据集,包含68,182个句子,超过100万个单词,涵盖从幼儿园到研究生院的19个可读性等级。该数据集由纽约大学阿布扎比计算语言模型实验室和Zayed大学阿拉伯语言研究中心共同创建,旨在为阿拉伯语文本复杂性评估提供全面的资源。数据集的构建过程包括手动标注,并确保了高水平的标注者间一致性。BAREC数据集的应用领域主要是文本可读性评估,目的是为了提高文学、语言学习和学术表现。
BAREC is a large-scale fine-grained Arabic readability assessment dataset containing 68,182 sentences and over one million words, covering 19 readability levels ranging from kindergarten to graduate school. This dataset was co-created by the Computational Language Model Lab at New York University Abu Dhabi and the Center for Arabic Language Studies at Zayed University, aiming to provide a comprehensive resource for Arabic text complexity assessment. The dataset construction process includes manual annotation and ensures a high level of inter-annotator agreement. The main application areas of the BAREC dataset are text readability assessment, with the goal of enhancing literature, language learning and academic performance.




