28th International Congress on project Management "Project Management in the Digital Decade"

Time overruns are a common occurrence in construction projects around the world. Identifying the causes of delays is the first and most important step in eliminating or mitigating delays on future construction projects. This research aims to propose a framework for the detection of the causes of delays in construction projects based on artificial intelligence techniques. Existing research is based on expert opinion, burdened by subjectivism and bias, and without using the potential of the text documentation from the project. Post-project analyses are a valuable source of information about delays. However, the complexity of coordinating the project team of the completed project contributes to the negative trend of producing these documents. The research presents the key elements of the delay detection system based on Transformer language models – DREAM (Delay Root causes Extraction and Analysis Model). The proposed DREAM combines expert knowledge, Transformer language models, and Minutes of weekly Meetings (MoMs) as substantial unstructured textual documentation. The results of the proposed system should enable experts to understand the nature of the causes of delay and initiate the process of intelligent decision-making on future projects.