A novel framework of knowledge transfer system for construction projects based on knowledge graph and transfer learning

Abstract

For construction enterprises, efficient knowledge sharing among projects not only effectively improves enterprise technology, level of management and competitiveness, but also promotes their sustainable development. Given the many benefits of knowledge management, enterprises have an urgent need for project knowledge sharing methods and tools. In this study, we build an automated and intelligent framework for a construction project knowledge transfer system based on knowledge graph and transfer learning. This framework aims to solve the problem of ineffective knowledge transfer that is encountered in the management of construction project knowledge sharing. First, to discover the relationship among knowledge and further obtain the relationships among projects, we design a domain knowledge graph ontology for construction projects and build an example. Then, based on this domain knowledge graph and combining construction data distance distribution with construction project knowledge background, we design a new construction project similarity measurement algorithm (PBG-MMD), which can guide the selection of the knowledge transfer source domain. Finally, a new transfer learning method is developed to automatically select the transfer source domain according to the domain context. The framework provides an effective answer for the problem of “what to transfer” in transfer learning and provides an effective solution to address the problem of “how to transfer” during knowledge transfer. Through the verification of practical case data, the proposed framework successfully realizes knowledge transfer among construction projects and provides an automated and intelligent knowledge sharing approach for construction enterprises.