CODE-ACCORD

CODE-ACCORD数据集是由伯明翰城市大学计算机、工程与建筑环境学院等机构合作创建，专注于建筑法规领域的自然语言处理和信息提取研究。该数据集包含从英国和芬兰建筑法规中提取的862条自包含句子，每条句子均经过实体和关系标注，以支持自动化合规检查。数据集的创建过程涉及从官方文档中提取文本，经过自动和手动过滤，确保句子的自包含性和准确性。CODE-ACCORD数据集主要应用于建筑、工程和施工（AEC）行业的自动化合规检查，旨在通过机器学习技术提高合规检查的效率和准确性。

The CODE-ACCORD dataset was collaboratively created by institutions including the School of Computing, Engineering and Built Environment at Birmingham City University, focusing on natural language processing and information extraction research in the field of building codes. This dataset contains 862 self-contained sentences extracted from the building codes of the United Kingdom and Finland, with each sentence annotated with entities and relations to support automated compliance checking. The dataset creation process involves extracting text from official documents, followed by automatic and manual filtering to ensure the self-containment and accuracy of the sentences. The CODE-ACCORD dataset is primarily applied to automated compliance checking in the Architecture, Engineering and Construction (AEC) industry, aiming to improve the efficiency and accuracy of compliance checking through machine learning technologies.