Retrieving similar cases for construction project risk management using Natural Language Processing techniques

Abstract

Abstract Case-based reasoning (CBR) is an important approach in construction project risk management. It emphasises that previous knowledge and experience of accidents and risks are highly valuable and could contribute to avoiding similar risks in new situations. In the CBR cycle, retrieving useful information is the first and the most important step. To facilitate the CBR for practical use, some researchers and organisations have established construction accident databases and their size is growing. However, as those documents are written in everyday language using different ways of expression, how information in similar cases is retrieved quickly and accurately from the database is still a huge challenge. In order to improve the efficiency and performance of risk case retrieval, this paper proposes an approach of combining the use of two Natural Language Processing (NLP) techniques, i.e. Vector Space Model (VSM) and semantic query expansion, and outlines a framework for this Risk Case Retrieval System. A prototype system is developed using the Python programming language to support the implementation of the proposed method. Preliminary test results show that the proposed system is capable of retrieving similar cases automatically and returning, for example, the top 10 similar cases.