67471 research outputs found
Sort by
Edge-Based Fault diagnosis of Autonomous Ground Robots with Simultaneous Actuator and Sensor Faults
No full textIn this paper, an edge-based fault diagnosis scheme is developed for a nonlinear ground vehicle model with possible occurrence of simultaneous actuator faults in the form of loss of effectiveness (LOE) and sensor bias faults. Based on the vehicle and fault models under consideration, the unknown fault parameters are estimated using online adaptive estimation methods. The estimated fault parameters can be utilized to detect, isolate, and accommodate faults in robotic system components, enabling the development of self-diagnostic smart sensors and actuators. Real-time experimental results using a ground robot are shown to illustrate the effectiveness of the proposed algorithm
Analyzing Data from Urban Citizen Participation by Applying the Retrieval Augmented Generation Architecture
No full textThis study explores the application of the retrieval-augmented generation architecture for large language models in analyzing citizens' contributions from urban participation. Existing literature highlights the potential of large language models to streamline analytical processes. However, challenges regarding required functions, domain expertise, and transparency remain underexplored. This research addresses these issues through a design science research approach. We identified eleven issues with a systematic literature review and twelve expert interviews, formulated twelve meta-requirements, and derived four design principles on which we developed a web prototype. We evaluated it with 42 experts from a crowdsourcing platform. Our findings demonstrate that retrieval-augmented generation models can enhance efficiency in automated categorization, sentiment analysis, and summarization by focusing the model's attention. However, transparency limitations persist as an ongoing challenge. Our findings contribute to existing knowledge by illustrating how hybrid intelligent systems can improve urban experts' ability to analyze and interpret participation data in smart cities
DeepSeek in China: AI Hiring or Bias Hiring?
No full textAlgorithmic hiring tools based on large language models (LLMs) are increasingly adopted, yet studies show that such systems replicate historical labor market biases. Prior research has largely focused on Western contexts, leaving limited understanding of how these issues manifest in China. This study evaluates DeepSeek, a leading Chinese LLM used in recruitment, to fill this gap. We combine linear regression with explainable machine learning techniques to quantify the influence of demographic and job-related factors on candidate scores. Results reveal systematic disparities, with applicants aged 35 and above, as well as female candidates receiving lower predicted scores. These findings highlight entrenched inequities in China’s labor market, provide a novel perspective on international implicit bias research, and demonstrate how combined methods reveal complex bias patterns. Beyond its academic contributions, the study offers practical guidance for fairness-aware AI deployment and contributes to ongoing discussions on trustworthy AI and regulation
When Image Emotion Outpaces Text: The Role of Emotional Intensity Incongruity in Driving Engagement on Instagram
No full textDrawing on Emotion As Social Information Theory, Cognitive Dissonance Theory, and the Elaboration Likelihood Model, we investigate how mismatched emotional intensity across visual and textual modalities impacts engagement. We analyze how the relative emotional dominance of the visual modality—measured as the raw difference in emotional intensity between image and text within a post—affects engagement. Our dataset includes 829,602 Instagram posts from 28,827 North American, English-speaking influencers, each annotated with probabilistic intensity scores across six basic emotions: anger, disgust, fear, joy, sadness, or surprise. The results demonstrate that when image-based emotional intensity exceeds textual expression—captured as the signed difference between the two—comment counts increase across most emotions, suggesting that engagement is sensitive not just to the magnitude, but to the direction of emotional imbalance. However, consistent with Cognitive Dissonance Theory, this positive effect diminishes and reverses beyond a critical threshold of emotional intensity incongruity between image and text
A User-Centric Taxonomy for Visualization-Oriented Natural Language Interfaces
Get PDFData analysis is an important success factor for many companies and organizations today. Faced with the challenges of ever-increasing data volumes and the lack of specially trained data analysts, the importance of user-friendly data visualization tools has increased. Natural language processing has enabled a more barrier free way of interacting with data in the form of visualization-oriented natural language interfaces (V-NLIs). In recent years various approaches of these systems have shown up in the literature. In this paper we present a methodically derived taxonomy that allows a classification of these approaches in terms of their user interaction. The results help scientists and practitioners to gain an overview of the field and assist them in making design decisions
Introduction to the Minitrack on Collaboration with Intelligent Systems: Machines as Teammates
No full textAn Amplitude-Based Threshold Design Process for Reliable RMS-Energy Oscillation Detectors
Get PDFOscillation detection is one of the key reliability-enhancing capabilities enabled by phasor measurement unit (PMU) technology. Several grid operators have deployed commercial platforms that implement oscillation detection by monitoring the RMS-energy of measured signals. Oscillations are detected when the signal's energy crosses a predetermined threshold. The current process of setting these thresholds requires the analysis of at least three months of historical data, which is time consuming and expensive. The thresholding process seeks to avoid false alarms, but it does not evaluate the likelihood that a given oscillation amplitude will trigger detection. As a result, thresholds often require re-tuning to avoid nuisance alarms. This paper proposes a method for designing RMS-energy thresholds that directly integrates oscillation amplitudes selected by the practitioner. The method improves upon previously published work by providing the practitioner with 1) guidance on the range of amplitudes that will lead to a successful design, and 2) a complete summary of the threshold's expected performance. The method is explained and tested using field-measured PMU data to demonstrate its practical viability
Implementation Barriers to Digital Mental Health Screening: A Qualitative Study of CAT-MH Implementations
Get PDFThe implementation of digital mental health screening tools shows promise for better identifying mental health needs in primary care settings, but their adoption remains challenging. This study uses CAT-MH as an example to investigate implementation barriers that prevent these tools from being integrated into clinical workflows across various healthcare settings. Following an exploratory qualitative design, we conducted interviews with eight healthcare professionals who have practical experience in implementing CAT-MH in their institutions. We used the EPIS framework to develop interview protocols and data analysis procedures. Our analysis revealed three primary barriers to implementation of CAT-MH: (1) EHR integration, (2) ambiguous role definitions and (3) insufficient organizational support. Our findings indicate that institutional-level coordination between stakeholders that influence system-wide implementation and clinical workflows must improve to support both effective integration and sustained use of innovative digital screening tool
Explainable AI in Content Moderation: Global, Local, and Narrative Approaches to Trust and Plausibility
Get PDFWe have explored various explainable AI scopes in fake news detection and found individuals trust AI when its explanation feels plausible. Drawing on the Heuristic–Systematic Model, we claim that plausibility mainly drives reliance. In an experiment, we compared seven increasingly detailed explanation formats—ranging from a bare flag to keyword lists, token-highlight views, and two levels of narrative—while controlling model accuracy. Trust did not rise monotonically with extra detail. Unedited token highlights lowered trust relative to the keyword list; pruning irrelevant tokens restored it; only the contextual narrative produced the highest trust. Mediation analysis confirmed the theory: the plausibility → trust path was β ≈ 0.55 (p < .001), and all indirect effects were significant. Our findings suggest that, in content-moderation settings, explanations that foreground plausibility through concise, context-anchored narratives can foster higher reviewer trust than simply exposing additional model internals
Geospatial Analysis of Wildfire Impact and Predictive Modeling of Susceptibility: A Case Study of Maui, Hawaii and California
No full textCatastrophic wildfires pose a growing global threat. This study analyzes their impacts through a comparative analysis, developing a predictive risk model that integrates satellite remote sensing within a cloud-based Geographic Information Systems (GIS) framework. On the Google Earth Engine platform, the differenced Normalized Burn Ratio (dNBR) and Normalized Difference Vegetation Index (NDVI) were applied to assess burn severity and vegetation health for the 2023 Maui wildfire and the early 2025 California wildfire. Despite stronger resilience, California's fires cause a heavier overall economic impact, while Maui's smaller fire delivered a more concentrated, catastrophic blow to its community. The predictive risk model demonstrated high accuracy when validated against historical fire data, successfully identifying low fuel moisture, topography, land cover and human factors as key drivers of susceptibility. This research underscores the need for context-specific management and shows that GIS and cloud-based analysis are powerful tools for enhancing wildfire resilience, response, and planning