Threat detection in smart homes: a sociotechnical multimodal conversational approach for improved cyber situational awareness. [Dataset]

Abstract

Smart homes are becoming increasingly more complex and difficult to defend. Expanding Internet of Things (IoT) devices has reshaped socio-technical interactions within smart homes, yet security remains a secondary concern. In addition, users have been shown to lack awareness of potential vulnerabilities, leaving smart homes susceptible to attacks. The file associated with this output contains supplementary images, text, TEX, BST, CLO and PDF files.

Citation

MCDERMOTT, C.D. and NICHO, M. 2025. Threat detection in smart homes: a sociotechnical multimodal conversational approach for improved cyber situational awareness. [Dataset]. International journal of information security [online], 24, article number 173. Available from: https://link.springer.com/article/10.1007/s10207-025-01051-x#Sec42

• Acceptance Date: May 20, 2025
• Online Publication Date: Jul 11, 2025
• Publication Date: Jul 11, 2025
• Deposit Date: Jul 25, 2025
• Publicly Available Date: Jul 25, 2025
• Publisher: Springer
• DOI: https://doi.org/10.1007/s10207-025-01051-x
• Keywords: Intrusion detection; Sociotechnical; Cyber situational awareness; Iot; Smart homes; Chatbot; Conversational agents; Multi-modal
• Public URL: https://rgu-repository.worktribe.com/output/2935209
• Publisher URL: https://link.springer.com/article/10.1007/s10207-025-01051-x#Sec42
• Related Public URLs: https://rgu-repository.worktribe.com/output/2842640 (Journal article)
• Type of Data: Images, text, TEX, BST, CLO and PDF files
• Collection Date: May 20, 2025
• Collection Method: The study employed a mixed-methods design, incorporating both quantitative and qualitative approaches, with pre-test and post-test measurements conducted over a 21-day period involving 16 participants. Participants were divided into two groups (Control and Intervention) and completed the study in a home environment rather than a lab. This methodology facilitated robust statistical analysis using Friedman repeated measures one-way and Wilcoxon signed-rank tests while maintaining control over variables to isolate agent-specific impacts. Additionally, the approach supported iterative design improvements, enabling the evaluation of usability, task efficiency, and user satisfaction across multiple metrics. The three levels of Endsley's model were used to calculate the mean Perception (pe), Comprehension (co), and Projection(pr). The data collected was ordinal and used as a dependent variable. In addition, given the sociotechnical focus on designing technology in a way that supports human needs, the usability of each agent was measured. This involved measuring the efficiency (in seconds) with which participants could assimilate information about events in their environment, synthesise this into a meaningful understanding of the situation, and the accuracy with which they could identify threats in a network. To evaluate the viability of using CAs to improve CSA a series of use-cases were developed, a common strategy in human-centered design. Each use-case was designed to represent a realistic example of how a user might use the agents to monitor smart device and network activity.