67471 research outputs found
Sort by
Data Governance Practices for Generative AI Powered Organizational Knowledge Management Systems Using Retrieval Augmented Generation
Get PDFThis study examines how data governance supports the success of generative AI-based Knowledge Management Systems (KMS) using Retrieval-Augmented Generation (RAG) in large enterprises. Drawing on a multi-case study methodology, the research identifies 17 distinct data governance practices and synthesises them into a conceptual framework that theorises their contribution to KMS success. The adoption of these practices is shaped by the dynamically evolving technological affordances of generative AI and RAG, as well as the contextual challenges posed by the predominantly ingested semi-structured and unstructured textual data. While the identified practices enable value-add, they also introduce strategic trade-offs, particularly in balancing data protection and expected benefits. This study contributes to the evolving discourse on data governance by extending its scope beyond structured data and highlighting its dynamic, context-sensitive role in AI-enabled KMS
Digital Participation and the Sociomaterial Effects of TikTok in a Municipal Context
Get PDFIn the context of digital participation in the public sector, the challenge of effectively utilizing digital initiatives to increase public engagement has emerged as a key area of research. In this study, we focus on the first municipality in Sweden to establish a TikTok channel, an initiative that seeks to foster a novel form of digital participation, providing young people with a platform to express their voices. Research on digital participation has traditionally explored how people access, use, and are affected by technology. Using a sociomaterial perspective, we explore digital participation and the intertwining of TikTok in everyday work in the municipality. This study contributes to research on the promise and complexity of digital participation, by demonstrating new dimensions through three sociomaterial effects, exposure, disruption, and trade-off. These insights also highlight why digital participation is not a straightforward solution that governments can simply implement without adopting a systemic approach
Design and Implementation of a Semiconductor Supply Chain Knowledge Graph for Developing a GraphRAG-Based Question Answering Model
Get PDFA knowledge graph (KG) supports effective knowledge extraction and management, driving active research in supply chain management (SCM). This study constructs a KG representing the semiconductor manufacturing supply chain and develops a GraphRAG-based question answering (QA) model. Representing such processes as a KG requires domain-specific ontology design, which is typically manual, time-consuming, and costly. To address this, we adapt an existing ontology from the automotive manufacturing domain to the semiconductor context. The QA model integrates the domain ontology into Chain-of-Thought (CoT) prompts to generate accurate responses to complex queries. This approach improves knowledge retrieval efficiency, mitigates the context-length limitations of a large language model (LLM), and enables interpretable answers by exposing the model’s reasoning process. We quantitatively evaluate the proposed GraphRAG model against the VectorRAG baseline, demonstrating superior performance in BERTScore and F1-score metrics. This framework highlights the benefits of combining ontological knowledge and CoT prompting for enhancing graph-based QA systems
Enhancing Fake News Detection Using GPT-2 with a Hybrid Deep Learning Approach
Get PDFNowadays, the spread of fake news represents a growing problem. To overcome this issue, it is essential to develop effective fake news detection systems. In this paper, we augment the process of fake news detection using GPT-2 alongside Convolutional Neural Networks (CNN) or Long Short-Term Memory (LSTM) networks. We use a dataset from Kaggle comprising of real and fake news articles. We compare the performance of the existing state-of-the-art real and fake news detection algorithms with our proposed hybrid model in Table 1. We blend GPT-2, known for contextual understanding, with either CNN or LSTM networks to capture more syntactical and semantic features from news articles and outperform the baseline algorithms. The preliminary results show that our hybrid model outperforms all the baseline algorithms on detecting real from fake news. By mixing generative pre-trained transformers with traditionally deep learning models, the robustness of misinformation detection systems can be significantly enhanced
Responsible AI by Design: Embedding Diversity, Equity, and Inclusion Values into AI Development Practices
Get PDFResponsible AI has emerged as a critical concern in Information Systems research, centered on embedding principles such as fairness, accountability, and transparency into the development of AI systems. The integration of these principles, particularly those of diversity, equity, and inclusion into everyday design practices remains inconsistent and under-theorized, leaving a gap between ethical aspiration and implementation. To address this, we draw on Value-Belief-Norm theory to model how personal DEI values translate into ethical development behavior. Using survey data from 194 AI professionals, we test a model where personal values influence awareness of consequences, moral responsibility, and personal norms, which in turn predicts the adoption of inclusive AI practices. All hypothesized relationships are supported, underscoring the importance of individual-level moral cognition in operationalizing RAI. This study advances IS scholarship by shifting the analytical focus from institutional governance to the cognitive and motivational mechanisms through which ethical AI is enacted in practice
Timed vs. Untimed Assessments in Computer Science Education: Mapping Student Engagement and Learning Progression Through Network Analysis
Get PDFAssessment plays a critical role in shaping learning outcomes in computer science education. This study applies network analysis to examine how different assessment designs—timed and untimed—affect student engagement and learning progression. Using data from an introductory programming course delivered in both online and in-person formats, we construct performance correlation networks to analyze similarities among students over time. Our findings show that untimed assessments, delivered via an interactive textbook platform, enable more individualized engagement patterns and clearer differentiation in learning outcomes. In contrast, timed assessments produce more standardized performance clusters, potentially masking underlying learning variability. Despite differences in modality, we observe consistent patterns across online and in-person settings, underscoring the influence of instructional design over delivery format. This work contributes to computing education by integrating learning analytics and pedagogical analysis, demonstrating how network-based methods can enhance assessment strategy evaluation and support more adaptive educational environments
Drawing the Line: Coping with Digital Stressors in Remote Work
Get PDFThe COVID-19 pandemic prompted a sudden and unprecedented shift to remote work, forcing organizations and employees to adapt to new working modalities. As hybrid work becomes the norm in the post-pandemic world, understanding the factors that support or hinder employee adjustment to remote work is crucial. This study investigates how knowledge workers adapt to remote work by examining the impact of two stressors—professional isolation and digital invasion—on remote work adjustment. Grounded in cognitive appraisal theory and the individual resilience literature, the study explores how boundary management strategies and planning alleviate these dysfunctional impacts. A field study was conducted among 389 employees in a multinational video game company operating globally. Results show that professional isolation and digital invasion are negatively associated with remote work adjustment. However, employees who adopt segmentation-based boundary strategies are better able to mitigate the negative effects of digital invasion. Moreover, planning positively contributes to remote work adjustment. This research offers both theoretical and practical insights into how individual coping strategies can facilitate successful adaptation in the evolving landscape of remote and hybrid work
