Digital Library of Gesellschaft für Informatik e.V.
Not a member yet
40758 research outputs found
Sort by
Impact of Preprocessing on Classification Results of Eye-Tracking-Data
No full textEye-tracking data provides valuable insights into human behavior, but its noisy and unstable nature necessitates robust preprocessing for accurate analysis. This study evaluates a tailored preprocessing pipeline designed to enhance machine learning classifier performance. Unlike prior research focusing on isolated preprocessing steps, this work systematically combines and compares techniques, including missing value imputation, outlier handling, and normalization, specifically optimized for eye-tracking data. The pipeline's impact is tested on classification accuracy, particularly in detecting academic dishonesty. By experimenting with diverse methods for handling missing data, outliers, and feature scaling, we assess their combined effects on classifier performance. A Random Forest classifier is utilized due to its proven effectiveness in prior studies \cite{nurwulan_random_2020}. This research not only builds on earlier findings but extends them by optimizing each preprocessing step. Results show a well-designed pipeline significantly enhances classification accuracy, offering insights into optimal preprocessing techniques for behavioral prediction tasks
A case study of the MEUSec method to enhance user experience and information security of digital identity wallets
No full textDigital identity wallets enable the storage and management of digital identities and verifiable credentials in one place on end users’ devices. This includes discount vouchers or customer cards, and security-critical data such as ID cards or driving licences. However, digital identity wallets face significant challenges due to weaknesses in user experience and information security. Users often find it difficult to understand the concept of digital identity wallets, resulting in personal information being inadvertently shared with untrusted parties. Additionally, user experience and information security can influence each other, so that both aspects must be evaluated and improved together. To this end, the Method for Enhancing User Experience and Information Security (MEUSec) can be used. This article reports on an experimental application of the MEUSec method to the wallet “Hidy” with two research goals: First, to evaluate the MEUSec method and the quality of its results against a set of criteria, and second, to collect suggestions for improving the user experience and information security of the Hidy wallet. In total, 41 weaknesses and 7 strengths of user experience and information security, 32 heuristics and 26 improvement suggestions for the Hidy wallet could be identified
ChatAnalysis revisited: can ChatGPT undermine privacy in smart homes with data analysis?
No full textLarge Language Models (LLMs) have demonstrated potential in automating data-driven tasks, enabling non-experts to analyze raw inputs such as tables or sensor data using conversational queries. Advances in Machine Learning (ML) and Human-Computer Interaction (HCI) have further reduced entry barriers, pairing sophisticated model capabilities and background knowledge with user-friendly interfaces like chatbots. While empowering users, this raises critical privacy concerns when used to analyze data from personal spaces, such as smart-home environments. This paper investigates the capabilities of LLMs, specifically GPT-4 and GPT-4o, in analyzing smart-home sensor data to infer human activities, unusual activities, and daily routines. We use datasets from the CASAS project, which include data from connected devices such as motion sensors, door sensors, lamps, and thermometers. Extending our prior work, we evaluate whether advances in model design, prompt engineering, and pre-trained knowledge enhance performance in these tasks and thus increase privacy risks. Our findings reveal that GPT-4 infers daily activities and unusual activities with some accuracy but struggles with daily routines. With our experimental setup, GPT-4o underperforms its predecessor, even when supported by structured CO-STAR prompts and labeled data. Both models exhibit extensive background knowledge about daily routines, underscoring the potential for privacy violations in smart-home contexts
Vorwort
No full textOver the past decade, the importance of data has increased in almost all scientific disciplines, e.g., in meteorology, genomics, physical simulations, biological and environmental research, medicine, and, more recently, in the humanities and social sciences. The workshop will focus on two main aspects: how tools can help achieve the FAIR (Findable, Accessible, Interoperable, Reusable) data principles, including legal aspects, and the scientific contributions that the database and information systems community can make to the development of Germany’s National Research Data Infrastructure (NFDI). The event aims to bring together scientists from various disciplines and NFDI consortia with database researchers to discuss practical issues in data science and the latest big data technologies
Exploring technical implications and design opportunities for interactive and engaging telepresence robots in rehabilitation – results from an ethnographic requirement analysis with patients and health-care professionals
No full textThis paper explores technical implications and design opportunities that are conceptualized to inform a socio-robotic system with digital applications to support the recovery process of patients within a rehabilitation facility. By conducting observations and interviews with patients and therapists, we identified key challenges and design opportunities in a specific orthopaedic rehabilitation context and process. The findings indicate the design potentials of a socio-robotic system to enhance patient engagement and recovery by providing personalized activities, a meaningful interaction and a motivating surrounding by using music-based exercises. Our research suggests that integrating digital applications with robotic systems may be used in the long-run to offer tailored exercises, stimulating concepts to motivate and maintain patients in therapy process, real-time feedback, and data-driven progress tracking, thereby improving the overall therapeutic outcomes. By addressing these factors, our proposed socio-robotic system aims to create a more interactive and engaging orthopaedic rehabilitation experience and environment, ultimately supporting patient recovery and improving overall treatment
Shift Right Testen – Eine Annäherung
No full textShift Right Testen bezeichnet das Testen im rechten Teil des DevOps-Prozesses, d.h. in den Phasen „Bereitstellen“, „Freigeben“, „Beobachten“ und „Lernen“. Der Arbeitskreis Innovative Testmethoden (ITM) erarbeitete einige Überlegungen zum Shift Right Testen, zu Use-Cases und 14 Aspekten
A Multi-level Reference Model and a Dedicated Method for Cyber-Security by Design
No full textThe increased reliance of organizations on information technology inherently increases their vulnerability to cyber-security attacks. As a response, a host of cyber-security approaches exists. While useful, these approaches exhibit shortcomings such as an inclination to be fragmented, not accounting for up-to-date organizational data, focusing on singular vulnerabilities only, and being reactive, i.e., focusing on patching up vulnerabilities in current systems. The paper presents and evaluates a modeling method aiming to address those shortcomings and to support security by design with a focus on the electricity sector. The proposed modeling method encompasses a multi-level reference model reconstructing and integrating existing initiatives and supporting top-down and bottom-up analyses. Compared to earlier work, the paper contributes (1) a process model for cyber-security by design, which proactively considers security as a first-class citizen during the design process, (2) a complete coverage of the multi-level model, in terms of three views complementing the introduced process model, (3) an elaborated evaluation, in terms of reporting on an additional design science cycle
Foveated Path Tracing with Configurable Sampling and Block-Based Rendering
No full textPath tracing offers high-fidelity rendering but remains impractical for real-time applications due to slow convergence and noise. We present a dynamic foveated path tracing technique that leverages visual perception by reducing sampling towards peripheral regions. Our system achieves up to 25-fold performance gains on complex scenes at resolution with minimal perceptual degradation. We validate its effectiveness using structured error maps across varying sampling rates and foveated region sizes, establishing a foundation for future research in perceptual photorealistic rendering
