67471 research outputs found
Sort by
Automatic Extraction of Protected Health Information from Multilingual Hacker Communities
Get PDFProtected Health Information (PHI, e.g., electronic health records, insurance information) is increasingly stolen in data breaches by malicious actors with the intent to sell to others in hacker communities. These actors often protect themselves by describing the content and availability of PHI data using encrypted messaging platforms (e.g., Telegram & Discord). However, the extent and nature of these PHI discussions are not well known. Therefore, in this research, we propose a Named Entity Recognition Framework for PHI (NERF-PHI) to systematically analyze PHI-related hacker conversations. To conduct our research, we collected more than three million multilingual hacker posts from Discord servers and Telegram groups. Utilizing open-source machine translation tools, we translated conversations to English and extracted information related to vulnerable individuals and medical entities. Results from our study suggest that encoder-based Large Language Models show significant promise for extracting PHI-related information from hacker communities and can be used by cybersecurity professionals and law enforcement to combat PHI misuse. Our study is also one of the first comprehensive analyses of multilingual PHI discussions in hacker communities
Integrating an Intelligent Tutoring Chatbot into Higher Education: The Case of Teaching Declarative Process Modeling through Generative AI
Get PDFThis paper presents a case study on the classroom use of an intelligent tutoring chatbot designed to support teaching Declarative Process Modeling within a master’s-level course. Replacing traditional lectures, the chatbot delivered interactive, task-based modules with real-time feedback and conversational support. Across three design iterations, we collected and analyzed student feedback, system usage data, and survey responses to examine how learners engaged with the chatbot, what challenges they encountered, and how their input informed system refinement. Students used the chatbot to self-regulate their learning, frequently asked clarifying questions, and responded critically to feedback. Their contributions shaped improvements in instructional clarity and usability. The findings demonstrate the potential of Generative AI to enable adaptive, student-centered learning experiences in higher education when thoughtfully integrated into course design
Inter-Organizational Collaborative Machine Learning: A Problem Space Exploration
Get PDFOrganizations can benefit substantially from machine learning (ML), but individual organizations often encounter resource constraints (e.g., related to data and computational resources) when using ML. Inter-organizational collaboration represents a promising approach to overcoming such resource constraints. While there are various inter-organizational collaborative ML solutions, we lack a thorough understanding of the corresponding problem space. Based on a structured literature review and qualitative content analysis, we explore the problem space of inter-organizational collaborative ML and uncover six stakeholders, six needs, seven goals, and seven requirements. We also identify four promising future research directions. Our study lays a foundation for developing design knowledge and more targeted solutions for inter-organizational collaborative ML
From Logs to Language and Vision: A Review on Integrating Multimodal Data into Predictive Process Monitoring
Get PDFPredictive Process Monitoring (PPM) forecasts future outcomes of business processes based on event logs. However, current approaches often overlook rich multimodal data – such as text, audio, video, and sensor inputs – that are increasingly generated in real-world settings. This paper presents a systematic literature review of 89 studies on multimodal data integration in PPM and related BPM tasks. We identify dominant modalities (e.g., sensor and text), methodological trends, and application domains, along with key challenges such as data fusion complexity, scarcity of labeled datasets, and model interpretability. At the same time, we highlight opportunities, including context-aware prediction, synthetic data augmentation, and real-time decision support. We propose future research directions which lay the foundation for advancing multimodal, intelligent BPM systems in complex environments
New Hires, New Worlds: Metaverse Onboarding and Its Impact on Organizational Socialization
Get PDFFrequent job changes and remote work have made effective onboarding more important—and more challenging—than ever before. As newcomers must quickly internalize organizational values, build relationships, and navigate unfamiliar roles, socialization becomes a critical factor in employee retention and performance. Traditional digital tools like videoconferencing often fall short in replicating key elements of in-person interaction, such as nonverbal cues, shared spatial context, and informal communication. In response, organizations are beginning to explore immersive technologies like the metaverse to bridge this gap. However, despite growing managerial interest, little is known about whether metaverse-based onboarding can meaningfully support newcomer socialization. Therefore, this study compares onboarding experiences in Zoom and a metaverse platform (Glue) through lab and online experiments. By investigating how immersive environments shape socialization processes, we offer theoretical and practical insights into designing more effective remote onboarding strategies for the future of work
Synthetic Faces in a Real Industry: How AI-Generated Deepfakes Challenge Actors’ Ownership of Likeness in the Film & TV Industry?
Get PDFDeepfake technologies are transforming screen industries by enabling realistic simulations of human performance. This study investigates how AI-generated deepfake audio and videos challenge actors’ ownership of their likeness, focusing on underrepresented perspectives from South Asia. While existing scholarship centres on celebrity cases or Western legal frameworks, this research draws on sixteen semi-structured interviews with actors in India, Pakistan, and Bangladesh. Using grounded theory analysis, it explores how synthetic replication endangers legal and economic rights, artistic integrity, consent, and emotional agency. Participants described exploitative contracts, loss of control over digital identity, and rising precarity, especially among lesser-known performers. Despite limited institutional support, actors expressed emerging forms of resistance and belief in the emotional depth and presence that only human performers can offer. By centring marginalised voices in AI discourse, this study offers a more grounded, ethically informed understanding of how synthetic media is reshaping labour and identity in film and television
Measuring Stakeholder Voting Coalition Stability in Regional Transmission Organizations
Get PDFWe use network analysis to probe votes taken at the top-level committees in three RTOs: the New England Independent System Operator, the New York Independent System Operator, and PJM. Whereas CAISO, MISO, and PJM were chosen because of the diversity in their stakeholder processes, ISO-NE, NYISO, and PJM were chosen because of their similarities, particularly their hierarchical voting structure and divisions of stakeholders into similar classes (i.e generators, transmission owners, end-use customers, and so on). This network approach allows us to look at how political power structures vary across the three RTOs based on the unique differences in the voting structures between the RTOs. We address three questions based on this network analysis: Whether specific constituencies can be identified; whether these constituencies are stable over different voting issues; and whether they are stable over time
Credibility Staining and the Boundaries of Expertise: The Reputational Cost of Commenting on Polarized Topics
Get PDFThis study introduces the concept of credibility staining, the reputational harm experts may experience when their commentary on polarizing issues undermines perceptions of their trustworthiness and expertise, even within their own domain. Such risks are increasingly relevant in today’s digital media environment, where experts often offer opinions on contentious topics outside their area of expertise. We tested this phenomenon using a 2x2 factorial experiment in which participants viewed a fictitious marine biologist’s tweets on both a domain-relevant issue (fishing zone expansion) and a domain-irrelevant, polarized issue (gun control). When the expert’s stance on gun control conflicted with participants’ beliefs, perceived credibility declined – even in their area of specialization. These effects were strongest among highly partisan individuals. The findings highlight how ideological alignment shapes credibility judgements and underscore the reputational risks of epistemic trespassing in digital discourse. We discuss implications for science communication and public trust in expertise
Between Promise and Practice: Challenges and Misperceptions of Applying Privacy Enhancing Technologies in Business Contexts
Get PDFApplying privacy-enhancing technologies (PETs), such as homomorphic encryption or differential privacy, promises to improve organizational cybersecurity strategies. However, in business contexts, significant gaps manifest between their technical capabilities and organizational perceptions, indicating a mismatch between promise and practice. This paper presents the first comprehensive meta-analysis of organizational PET perceptions through a systematic review of 34 empirical studies. Our findings reveal that while regulatory pressures and reputational considerations drive adoption, organizations face substantial practical challenges, including complexity management and insufficient understanding of technological capabilities. Even experienced practitioners show misperceptions about PET functionality, leading to misconfigurations that undermine promised privacy benefits. Thus, misperceptions directly impact cybersecurity effectiveness, as organizations may overestimate deployed protections or underutilize available capabilities. Consequently, our analysis highlights the need for and recommends implementing improved education, regular reassessments of current beliefs regarding PETs, and transparency mechanisms to translate potential into successful enterprise cybersecurity