Mining LDA topics on construction engineering change risks based on graded evidence

• Published in: PloS one 2024-06, Vol.19 (6), p.e0303424
• Main Authors: Jin, Lianghai, Li, Chenxi, Zhu, Zhongrong, Zou, Songxiang, Sun, Xushu
• Format: Article
• Language: English
• Subjects: Biology and Life Sciences; Computational linguistics; Computer and Information Sciences; Construction accidents & safety; Construction engineering; Construction Industry; Data mining; Data Mining - methods; Decision making; Design defects; Design engineering; Dirichlet problem; Engineering; Engineering and Technology; Hazard identification; Humans; Language processing; Medicine and Health Sciences; Methods; Natural language interfaces; Reliability; Risk assessment; Risk Assessment - methods; Risk management; Sentiment analysis; Social Sciences; Surveys and Questionnaires; Text analysis
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0303424

Engineering change (EC) risk may negatively impact project schedule, cost, quality, and stakeholder satisfaction. However, existing methods for managing EC risk have certain shortcomings in evidence selection and do not adequately consider the quality and reliability of evidence associated with EC risks. Evidence grading plays a crucial role in ensuring the reliability of decisions related to EC risks and can provide essential scientific and reliability support for decision-making. In order to explore the potential risks associated with architectural engineering changes (ECs) and identify the most significant ones, this study proposed a methodology that combines evidence grading theory and Latent Dirichlet Allocation (LDA) topic analysis means. Initially, the evidence-based grading theory served as the creation of a grading table for evidence sources related to EC risk. Specifically, we categorized the evidence sources into three levels based on their credibility. Subsequently, we selected evidence with higher credibility levels for textual analysis, utilizing the LDA topic model. This involved analyzing regulations, industry standards, and judgment documents related to EC, ultimately identifying the themes associated with EC risks. In addition, by combining EC risk topics with relevant literature, we identified factors influencing EC risks. Subsequently, we designed an expert survey questionnaire to determine the key risks and important risk topics associated with potential risks. The results show that by synthesizing information from both Class A and B evidence, a total of five prominent risk themes were identified, namely contract, technology, funds, personnel, and other hazards. Among them, the technical risk has the highest value, so it implies that the risk is the most important, and the key risks are engineering design defects, errors, and omissions.