| dc.description.abstract | Artificial Intelligence (AI) has revolutionized various industries in recent years by offering solutions to complex problems and enhancing decision-making processes. The integration of AI into decision support systems (DSSs) represents a significant advancement in sectors like healthcare, banking, governance, and search and rescue (SAR) operations. SAR missions, known for their complexity and the need for prompt and accurate decision-making, are a prime example of how AI can make a substantial difference.
The motivation behind this thesis was recognizing the growing importance of intelligent decision support systems that incorporate AI, also known as AI-informed DSSs, in SAR operations, as indicated by SAR actors in Norway. AI-informed DSSs have demonstrated impressive capabilities in processing vast amounts of unstructured data, filtering out irrelevant information, and generating practical recommendations since their inception in the 1970s. This study explores the roles of these systems in the context of SAR operations, which inherently involve the complex coordination of various organizations at multiple stages. Due to the complexities involved, highly knowledgeable stakeholders are necessary to locate and rescue people in distress.
The primary objective of this thesis is to investigate the incorporation of AI into the SAR DSS, with a particular emphasis on identifying the stages of SAR where AI integration could be most beneficial. Our work began with the investigation of current AI technologies and their potential to improve SAR DSS, particularly in terms of effective data management and planning. The following research questions have guided this PhD process. Although the scope of this PhD journey is Norwegian SAR operations, the questions are intended to develop extensive understanding and learning rather than simply test predefined hypotheses for region-specific usage. The questions are:
1. How is AI currently used to support decision-making in SAR operations?
2. How can AI be used to support decision-making in SAR operations?
2.1 How can AI be used to leverage past experiences in a decision support system for search and rescue planning and resource allocation?
2.2 How to design case-based and concept-based search services to improve the effectiveness of SAR planning?
The thesis consists of four research papers, each of which addresses different aspects related to SAR operations. Together, they paint an overall picture of the potential for improving SAR DSS through the use of AI technologies, specifically using knowledge graphs and case-based reasoning (CBR). The studies reported in these papers used multiple case studies, based on data gathered through expert interviews, document reviews, and observations. Data analysis consisted of both quantitative and qualitative analysis and summaries of findings. A design science research (DSR) approach, which is well-suited for information systems research, is used to address these research questions. This method is divided into five stages, which are problem awareness, suggestion, development, evaluation/validation, and conclusion.
The first research question (RQ1) is addressed in Research Paper I. This paper is based on a systematic literature review and bibliometric mapping. The main contribution of this paper is the identification of the research gap in the investigated domain. Moreover, an analysis of existing literature on search and rescue processes that use decision support systems, data management solutions, and artificial intelligence technologies is provided. The potential for knowledge transfer between application areas is discussed in this research paper. It has been observed that there is very little focus on SAR at sea, and the literature with more interactive data management solutions is most focused on land rescue. These findings strengthen the scope of overall PhD research.
The second research question (RQ2) and its subsequent questions (RQ2.1 and RQ2.2) are addressed in Research Paper III and Research Paper IV. These papers focus on How can AI be used in SAR operations? The data for these papers is collected from expert interviews, documentation reviews, and observations. The data collection through expert interviews resulted in Research Paper II which discusses the significance of experts’ involvement in research and their challenges and benefits. Based on the data analysis gathered from multiple data collection methods, a comprehensive set of requirements is established. To address these requirements, an architectural framework for SAR DSS is designed and developed. The framework is then evaluated using qualitative analysis. Additionally, a demonstrator system is implemented that provides the retrieval services and knowledge-base creation services presented in the framework. The knowledge base creation services use knowledge graphs to create a knowledge base with the data presented in the form of concepts. The retrieval services employ natural language processing (NLP) techniques, CBR, and concept-based retrieval for improving decision support in SAR operations. Both, qualitative and quantitative methods are used for the evaluation of a demonstrator system.
This thesis makes several contributions through its exploratory, descriptive, and explanatory research. Firstly, the exploratory and descriptive research identifies research gaps in the existing literature. Secondly, it highlights the importance of expert interviews as a method of data generation. The explanatory research establishes a set of requirements for improving SAR operations and designs an architectural framework for SAR DSS. Finally, the implementation of the demonstrator system based on the framework represents a significant step towards AI-informed decision support in SAR operations. The framework combines AI technology with the expertise of SAR professionals to enhance the efficiency and effectiveness of SAR operations.
This research sets the foundation for future integration of AI into SAR DSS, demonstrating the potential improvements in decision-making processes that this integration can bring. Additionally, the findings from this thesis contribute to the academic understanding of AI in decision support and suggest practical applications that can have a significant impact on SAR operations. These findings also indicate a promising direction for future research and development in AI-informed decision support for SAR, allowing for further exploration and implementation in this critical field. | en_US |
