Utilizing Topic Model To Review And Predict The Trends Of Construction Engineering And Management Study

Every year there are a lot of research works conducted, either completed or in progress in the Construction Engineering and Management(CEM) research field. Researchers in the CEM field are very interested to know what the future development trends is and hope able to conduct in-depth research based on the development trends. Therefore, a statistical analysis and review was carry out on CEM research trend(2001-2013), predict the future CEM prospects and understanding the Cross Topic Relationships(CTR) and analyze of its trends.The data sources are collected from 8 selected international journals(2001-2013). The data to be collected from each paper are its’ “Title” and “Abstract”. Due to the large amount of information involved, specialized software tools were developed and utilized for automatic processing, statistic and classify these data effectively. Construction Engineering and Management Body(CEMB) was constituted at the same time of data collection, for classify the data accordingly and more reasonable. In order to enhance the reliability and accuracy of the data, the data require to undergo data-preprocessing and verification. No exception for CEMB, Delphi Method were implemented. After data preprocessing and verification, Topic Model – Latent Dirichlet Allocation(LDA) algorithm were utilized for the statistics and data classification. A type of Markov Chain Monte Carlo(MCMC) algorithm, Gibbs sampling is used to resolve the LDA model.After data classification, labeling process needed to conduct manually according to CEMB. Verification needed execute again, result show that by using this method is able to achieved 76-93% of accuracy of result. This method is not only show the research trends from past 13 years, but also able to predict the potential topics based on the trends. Based on result, “Labor and Personnel Issue” and “Standard, Safety and Risk” are high potential development topic in future. Compared with existing similar research method – Bibliometrics, this method able to figure out the CTR between topics. Result show that, “technology”, “organization and management issue”, and “research method” accounted important in every topics of CEM, because it can help the productivity of construction industry.The result show that, there is more and more CEM papers were called, which mean more and more important CEM to society and the environment. Stakeholders now expect Value-For-Money(VFM) for construction activities and look towards new technologies, new management methods and control practices to reduce overall costs and accelerate construction works and with minimal social and environmental impact.