Subjectivity in data extraction - a study based on construction hazard identification

Simon Smith, P. Beausang, D. Moriarty, J. Campbell

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract / Description of output

The importance of effective mitigation of construction safety risks can be readily appreciated. However, the methods that are available that can optimise this process are not obvious and no standard procedure is laid down in the construction industry. An IT tool has been developed that allows mitigation measures to be identified for new risks. These measures are based on historically successful measures from a range of projects. Where more than one suitable mitigation technique is available the tool uses Case Based Reasoning to suggest the most appropriate. The data for this tool is based upon that contained in Method Statements, the industry standard format for task and safety planning and one of the more difficult aspects of the research is how to effectively extract the relevant data from these statements. Unfortunately the problem is compounded by the fact that the person extracting the data may be doing so in a subjective manner thus leading to issues with consistency and accuracy. A brief overview of the research work up to the present time will be presented, including an overview of the source of data used in this study, namely Method Statements. A discussion of the problem of subjectivity in data extraction will be provided. The paper will then present a subjectivity filter developed as a partial solution to this problem, together with initial results and findings when applying this filter to a real data extraction problem. These initial results have shown that it is possible to normalise data extraction when this exercise is carried out by multiple extractors.
Original languageEnglish
Title of host publicationARCOM 24th Annual Conference, Cardiff, September 1-3
Subtitle of host publicationAssociation of Researchers in Construction Management
Pages1065–73
Volume2
Publication statusPublished - 2008
Event24th Annual ARCOM Conference - Cardiff, United Kingdom
Duration: 1 Sept 20083 Sept 2008

Conference

Conference24th Annual ARCOM Conference
Country/TerritoryUnited Kingdom
CityCardiff
Period1/09/083/09/08

Fingerprint

Dive into the research topics of 'Subjectivity in data extraction - a study based on construction hazard identification'. Together they form a unique fingerprint.

Cite this