USING DIGITAL TECHNOLOGIES TO FACILITATE IDENTIFICATION OF POLITICAL RISKS IN INTERNATIONAL CONSTRUCTION PROJECTS

Date

2024-9-06

Author

Özyurt Ersöz, Beste

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

41
views

0
downloads

Cite This

In the context of international construction projects, effective and timely country risk assessment, particularly assessment of political risks is crucial due to uncertainty regarding socio-political conditions that have considerable impact on construction. Traditional risk identification and assessment methods, which rely on manual efforts and subjective evaluations, often fall short in capturing the complexities of political risk assessment. This research aims to address these challenges by leveraging digital technologies, particularly machine learning and natural language processing (NLP), to facilitate, enhance and to automate the political risk identification process. Based on findings of a needs analysis of construction professionals, a comprehensive taxonomy of political risk-related factors is developed, facilitating the systematic identification and categorization of risks specific to the construction industry and construction of domain ontology. The study explores the application of web crawling and AI-driven tools, such as ChatGPT, to automate the extraction of relevant political information from the web sources. Through a workshop and practical implementations, the research evaluates the effectiveness of these technologies in improving risk identification and assessment practices for contractors operating internationally. Findings demonstrate that web crawling gathers structured data from specific sources for targeted insights, as demonstrated by its application in extracting country-specific indicators from websites. While ChatGPT’s current performance in generating visuals and region-specific examples is limited and its knowledge base may not always reflect country-specific nuances, it provides broader insights and, demonstrates a promising understanding of political risk concepts and adapts well to scenario modifications, showing potential for enhancing political risk identification and assessment in the future. This suggests that with further development, particularly through a domain-specific large language model (LLM), it could become a valuable support tool for scenario-based political risk assessments. Finally, a framework for a domain-specific LLM that represents a significant advancement in automating political risk assessments is proposed. The findings contribute to the ongoing efforts to automate and refine the political risk identification and assessment processes, offering significant potential benefits for enhancing the strategic decisionmaking capabilities of construction firms in global markets.

Subject Keywords

International Construction, Political Risk Identification, Automation, Generative AI, Large Language Models

URI

https://hdl.handle.net/11511/111427

Collections

Graduate School of Natural and Applied Sciences, Thesis