Automated qualitative rule extraction based on bidirectional long shortterm memory model

Zhu, Xiaofeng, Li, Haijiang ORCID: https://orcid.org/0000-0001-6326-8133, Xiong, Guanyu and Song, Honghong 2022. Automated qualitative rule extraction based on bidirectional long shortterm memory model. Presented at: 29th EG-ICE International Workshop on Intelligent Computing in Engineering 2022, Aarhus, Denmark, 06-08 July 2022.

Preview

PDF - Accepted Post-Print Version Download (215kB) | Preview

Abstract

Digital transformation in the construction industry demands smart compliance checking against relevant standards to ensure high-quality project delivery. Due to the diverse characteristics, the qualitative rule extraction for standards remains labour intensive. Therefore, an efficient and automated rule extraction method is pivotal. The artificial neural network has been widely used for textual feature extraction in recent years. In this paper, the authors construct an automated rule extractor based on a bidirectional Long short-term memory (LSTM) neural network model, which can automate the extraction of qualitative rules in textual standards and achieves an accuracy of 96.5% in actual tests. The automated rule extractor can greatly improve the efficiency of converting unstructured textual rules to structured data. This approach can establish the basis for knowledge mining of qualitative standards as well as the development of large-scale compliance checking systems.

Item Type:	Conference or Workshop Item (Paper)
Status:	In Press
Schools:	Engineering
Date of First Compliant Deposit:	12 April 2022
Date of Acceptance:	28 March 2022
Last Modified:	30 Nov 2022 07:30
URI:	https://orca.cardiff.ac.uk/id/eprint/149027

Actions (repository staff only)

Edit Item