A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Construction Procedural Information Extraction from Textual Sources to Support Scheduling
Nowadays, construction schedule generation still mostly relies on manual efforts to arrange activity information and fulfill the scheduling tasks. There is a lack of automation and efficiency. The heavy reliance on manual efforts may also introduce human errors in the process. Although there are many construction scheduling software platforms, such as Synchro, Primavera 6, and Powerproject, they require manual input of activity information based on procedural documents or experience. To bridge the gap of automation in generating scheduling information, in this paper, the authors proposed a semantic natural language processing (NLP)-based information extraction (IE) method to automatically extract, process, and analyze procedural information from procedural documents to support construction scheduling applications. This method automatically generates a report of all activities in a right sequence considering each activity’s predecessors and successors, which can reduce manual efforts in arranging activity information for scheduling tasks. An experiment was conducted on a set of open-source specifications to demonstrate the performance of procedural information processing when using the proposed IE method. Comparing to a manually developed gold standard, 95.83% precision and 90.45% recall were achieved using the proposed IE method for the extraction of construction procedural information. In addition, running the IE algorithm to generate activity information led to 89.33% time saving when compared to manual activity information generation. The high performance achieved in accuracy and efficiency using the proposed method showed it is promising in bringing the traditional manual construction scheduling process one step closer to full automation.
Construction Procedural Information Extraction from Textual Sources to Support Scheduling
Nowadays, construction schedule generation still mostly relies on manual efforts to arrange activity information and fulfill the scheduling tasks. There is a lack of automation and efficiency. The heavy reliance on manual efforts may also introduce human errors in the process. Although there are many construction scheduling software platforms, such as Synchro, Primavera 6, and Powerproject, they require manual input of activity information based on procedural documents or experience. To bridge the gap of automation in generating scheduling information, in this paper, the authors proposed a semantic natural language processing (NLP)-based information extraction (IE) method to automatically extract, process, and analyze procedural information from procedural documents to support construction scheduling applications. This method automatically generates a report of all activities in a right sequence considering each activity’s predecessors and successors, which can reduce manual efforts in arranging activity information for scheduling tasks. An experiment was conducted on a set of open-source specifications to demonstrate the performance of procedural information processing when using the proposed IE method. Comparing to a manually developed gold standard, 95.83% precision and 90.45% recall were achieved using the proposed IE method for the extraction of construction procedural information. In addition, running the IE algorithm to generate activity information led to 89.33% time saving when compared to manual activity information generation. The high performance achieved in accuracy and efficiency using the proposed method showed it is promising in bringing the traditional manual construction scheduling process one step closer to full automation.
Construction Procedural Information Extraction from Textual Sources to Support Scheduling
Ren, Ran (author) / Zhang, Jiansong (author)
Construction Research Congress 2022 ; 2022 ; Arlington, Virginia
Construction Research Congress 2022 ; 330-339
2022-03-07
Conference paper
Electronic Resource
English
| British Library Conference Proceedings | 2012