proposition-level alignment dataset

本数据集名为“proposition-level alignment dataset”，由巴伊兰大学创建，旨在通过精确的命题级别对齐来优化摘要生成中的信息提取。数据集包含23,492条对齐实例，这些实例是从多文档摘要（MDS）评估数据中自动派生而来，用于训练监督对齐基线模型。创建过程中，采用了复杂的众包方法和高质量的开发与测试数据集。该数据集主要应用于文本摘要领域，特别是解决信息提取和摘要生成中的准确性问题。

This dataset, titled 'proposition-level alignment dataset', was developed by Bar-Ilan University with the goal of optimizing information extraction in summarization tasks through precise proposition-level alignment. It comprises 23,492 alignment instances automatically derived from multi-document summarization (MDS) evaluation data, and is utilized for training supervised alignment baseline models. During the dataset's creation, sophisticated crowdsourcing methodologies alongside high-quality development and test datasets were employed. This dataset is primarily applied in the domain of text summarization, specifically to resolve accuracy-related challenges in information extraction and summarization generation.