Data-driven and production-oriented tendering design using artificial intelligence

Abstract: Construction projects are facing an increase in requirements since the projects are getting larger, more technology is integrated into the buildings, and new sustainability and CO2 equivalent emissions requirements are introduced. As a result, requirement management quickly gets overwhelming, and instead of having systematic requirement management, the construction industry tends to trust craftsmanship. One method for a more systematic requirement management approach successful in other industries is the systems engineering approach, focusing on requirement decomposition and linking proper verifications and validations. This research project explores if a systems engineering approach, supported by natural language processing techniques, can enable more systematic requirement management in construction projects and facilitate knowledge transfer from completed projects to new tendering projects. The first part of the project explores how project requirements can be extracted, digitised, and analysed in an automated way and how this can benefit the tendering specialists. The study is conducted by first developing a work support tool targeting tendering specialists and then evaluating the challenges and benefits of such a tool through a workshop and surveys. The second part of the project explores inspection data generated in production software as a requirement and quality verification method. First, a dataset containing over 95000 production issues is examined to understand the data quality level of standardisation. Second, a survey addressing production specialists evaluates the current benefits of digital inspection reporting. Third, future benefits of using inspection data for knowledge transfers are explored by applying the Knowledge Discovery in Databases method and clustering techniques. The results show that applying natural language processing techniques can be a helpful tool for analysing construction project requirements, facilitating the identification of essential requirements, and enabling benchmarking between projects. The results from the clustering process suggested in this thesis show that inspection data can be used as a knowledge base for future projects and quality improvement within a project-based organisation. However, higher data quality and standardisation would benefit the knowledge-generation process. This research project provides insights into how artificial intelligence can facilitate knowledge transfer, enable data-informed design choices in tendering projects, and automate the requirements analysis in construction projects as a possible step towards more systematic requirements management.