A Naïve Bayes approach for ews detection by text mining of unstructured data: a construction project case

Knowledge Management became the focus of scientific study during the second half of the 20th century. Unstructured data in project management documents hold critical and time-based information about how the project is going-on. However, retrieving useful information from this data is still challenging task. The proposed approach presents a Naïve Bayes text mining approach to identify early warnings of failure in a project lifecycle in advance. The paper focused on a construction industry project to analyse the effectiveness of this approach. Hence, the primary focus of this research was to identify the lack of various project management aspects. The implemented system scanned critical management documents including Minutes of meeting to identify these so-called early warnings. The technique was evaluated against unseen Meeting Minutes PDF files labelled via expert feedback. The system reported 80.43% of the identified early warnings to be lack of onsite materials followed by lack of keen commitment to project milestones and lack of staff skills and training.