Capturing Implicit Structures in Unstructured Content of Construction Documents

In a paper-based paradigm, whether construction information is presented in structured or unstructured format is not a critical issue because human beings are good at processing both types of information. However, the paradigm shift from using paper-based media to electronic media requires better handling of unstructured information, as a computer is more adept at processing structured information. It is observed by the writers that many existing information systems in the construction industry simply mimic paper-based construction documents, not just in presentations but also in data structures, such as request for information (RFI) forms developed by many existing software vendors. The writers thus question whether enough attention has been put into understanding the impact of the paradigm shift on data management for construction documents. This paper argues that a construction document such as a RFI is usually used by construction professionals to serve particular business purposes, which restricts the content of such a document. Consequently, some types of unstructured information make references implicitly or explicitly to certain types of information that are often well structured such as building or project management information. Further, the occurrence of such cross references is constant and stable among different instances of the same type of documents. A major challenge is to properly capture, represent, and utilize such knowledge in order to improve information processing for construction professionals. Text analysis and statistical analysis are applied by this study to determine the patterns of the occurrences for certain types of concepts that are related to building components, contract documents, or project management in unstructured content. With the identified patterns, this paper further discusses the potential use of metadata models as a method to support processing unstructured content in those documents.