.png){kind=icon}
.png){kind=icon}
Return
Construction risk identification using a multi-sentence context-aware method
.png){kind=icon}
PRE
AI
DOI：10.1016/j.autcon.2024.105466.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Knowledge of risk events with potentially negative consequences from previous projects is essential for risk identification in early stages of new infrastructure projects. However, historical risk events are usually scattered in various sources and reports, rendering collecting such risk information time-consuming and expensive. To expand the current risk data sources and facilitate risk events' extraction, the paper presents a synthetic approach that utilizes Natural Language Processing (NLP) techniques to automatically identify and extract risk-related sentences from news articles. A supervised Multi-sentence Context-aware Risk Identification (MCRI) model is devised to exploit both sentence-level and multi-sentence level context to boost the sentence classification performance. The MCRI model outperformed several baseline models with a risk-class F1-score of 87.1% and an accuracy of 86.7%. This paper provides a baseline for future studies aimed at automating the extraction of project-level risk information within the construction domain.
Keywords:
Project -level risk
Risk identification
Context -aware text classification
Natural language processing
Journal
IF:
11.5
Papers:
6.2K
Citations:
4.2W