Automatic Compliance Checking (ACC) within the Architecture, Engineering, and Construction (AEC) sector necessitates automating the interpretation of building regulations to achieve its full potential. However, extracting information from textual rules to convert them to a machine-readable format has been a challenge due to the complexities associated with natural language and the limited resources that can support advanced machine-learning techniques. To address this challenge, we introduce CODE-ACCORD, a unique dataset compiled under the EU Horizon ACCORD project. CODE-ACCORD comprises 862 self-contained sentences extracted from the building regulations of England and Finland. Aligned with our core objective of facilitating information extraction from text for machine-readable rule generation, each sentence was annotated with entities and relations. Entities represent specific components such as "window" and "smoke detectors", while relations denote semantic associations between these entities, collectively capturing the conveyed ideas in natural language. We manually annotated all the sentences using a group of 12 annotators. Each sentence underwent annotations by multiple annotators and subsequently careful data curation to finalise annotations, ensuring their accuracy and reliability, thereby establishing the dataset as a solid ground truth. CODE-ACCORD offers a rich resource for diverse machine learning and natural language processing (NLP) related tasks in ACC, including text classification, entity recognition and relation extraction. To the best of our knowledge, this is the first entity and relation-annotated dataset in compliance checking, which is also publicly available.
In this paper we dealt with the comparison and linking between lexical resources with domain knowledge provided by ontologies. It is one of the issues for the combination of the Semantic Web Ontologies and Text Mining. We investigated the relations between the linguistics oriented and domain-specific semantics, by associating the GO biological process concepts to the FrameNet semantic frames. The result shows the gaps between the linguistics-oriented and domain-specific semantics on the classification of events and the grouping of target words. The result provides valuable information for the improvement of domain ontologies supporting for text mining systems. And also, it will result in benefits to language understanding technology.