Fakultät Informatik
Refine
Year of publication
Document Type
- conference proceeding (article) (171)
- Article (32)
- Part of a Book (20)
- Report (9)
- Bachelor-/Diplom Thesis (3)
- Book (3)
- Other (1)
Reviewed
- Begutachtet/Reviewed (201)
Keywords
- pathological speech (4)
- KI (3)
- Künstliche Intelligenz (3)
- neuropsychological tests (3)
- Artificial Intelligence (2)
- Automotive Security (2)
- ChatBot (2)
- Computerunterstützte Kommunikation (2)
- Graphdatenbank (2)
- Graphdatenmodell (2)
Institute
- Fakultät Informatik (239)
- Zentrum für Künstliche Intelligenz (KIZ) (55)
- Institut für e-Beratung (12)
- Fakultät Sozialwissenschaften (10)
- Fakultät Betriebswirtschaft (1)
- Fakultät Maschinenbau und Versorgungstechnik (1)
- Fakultät für Elektrotechnik Feinwerktechnik Informationstechnik (1)
- Institut für Angewandte Informatik (1)
- Lehr- und Kompetenzentwicklung (1)
Rising global demand for mental health support creates significant service delivery challenges, with asynchronous email counselling serving as a crucial low-threshold channel for accessing care. This paper presents CAIA, a co-designed AIbased tool suite that demonstrates responsible AI integration into counselling practice through seven LLM-driven functions enhanced by retrieval-augmented generation. A field evaluation involved 34 professional counsellors conducting authentic sessions with trained student counsellees (36 threads, 321 messages, 1,257 AI outputs). User behaviour analysis confirms substantial adoption, revealing that professional autonomy and information accuracy are decisive for sustained acceptance, with counsellors particularly valuing interpretive functionalities that provide new perspectives and stimulate professional reflection.
The expanding field of psychosocial online email counselling has created a demand for advanced tools to support counsellors in managing diverse client interactions. This paper introduces CAIA (Counsellor Artificial Intelligence Assistant), a comprehensive AI-driven system currently under active development, designed to condense and present information to counsellors in a structured manner. CAIA integrates several AI-based features, including Automatic Subject Generation, Case Summarisation, Timeline Management, Social Network Graph creation, and Case Classification. The features are designed to assist the counsellors, allowing them to engage more deeply with each case while ensuring that the client's needs remain paramount. Each feature presented in the paper addresses a specific problem in psychosocial online counselling, proposes a solution, discusses a technical implementation approach, and highlights the challenges encountered. Additionally, the paper discusses ethical considerations associated with an AI integration.
Large Language Models (LLMs) have the potential to enhance e-mental health and psychosocial e-mail counselling by automating tasks such as generating concise and relevant subject lines for client communications. However, concerns regarding accuracy, reliability, data privacy and resource efficiency persist. This study investigates the performance of several LLMs in generating subject lines for e-mail threads, yielding a total of 253 generated subjects. Each subject line was assessed by six raters, including five counselling professionals and one AI system, using a three-category quality scale (Good, Fair, Poor). The results show that LLMs can generally produce concise subject lines considered helpful by experts. While GPT-4o and GPT-3.5 Turbo outperformed other models, their use is restricted in mental health settings due to data protection concerns, making the evaluation of open-source models crucial. Among open-source models, SauerkrautLM LLama 3 70b (4-bit) and SauerkrautLM Mixtral 8x7b (both 8-bit and 4-bit versions) delivered promising results with potential for further development. In contrast, models with lower parameter counts produced predominantly poor outputs.
Deploying large language models (LLMs) in real-world applications requires robust safety guard models to detect and block harmful user prompts. While large safety guard models achieve strong performance, their computational
cost is substantial. To mitigate this, smaller distilled models are used, but they often underperform on “hard” examples where the larger model provides accurate predictions. We observe that many inputs can be reliablyhandled by the smaller model, while only a small fraction require the larger model’s capacity. Motivated by this, we propose SafeRoute, a binary router that distinguishes hard examples from easy ones. Our method selectively applies the larger safety guard model to the data that the router considers hard, improving efficiency while maintaining accuracy compared to solely using the larger safety guard model. Experimental results on multiple benchmark datasets demonstrate that our adaptive model selection significantly enhances the trade-off between computational cost and safety performance, outperforming relevant baselines.
Current work on speech-based dementia assessment focuses on either feature extraction to predict assessment scales, or on the automation of existing test procedures. Most research uses public data unquestioningly and rarely performs a detailed error analysis, focusing primarily on numerical performance. We perform an in-depth analysis of an automated standardized dementia assessment, the Syndrom-Kurz-Test. We find that while there is a high overall correlation with human annotators, due to certain artifacts, we observe high correlations for the severely impaired individuals, which is less true for the healthy or mildly impaired ones. Speech production decreases with cognitive decline, leading to overoptimistic correlations when test scoring
relies on word naming. Depending on the test design, fallback handling introduces further biases that favor certain groups. These pitfalls remain independent of group distributions in datasets and require differentiated analysis of target groups.
The Security Abstraction Model (SAM) is a meta-model that enables comprehensive security modeling and does not only consider technological aspects, but also takes into consideration the influence of social engineering. This plays an important role by exploiting vulnerabilities in human behavior and thus represents a dimension that must not be neglected in possible attacks on vehicle systems. This work is dedicated to the advancement of the Security Abstraction Model (SAM) with the aim of enabling improved modelling and analysis of vulnerabilities in the context of software and system security; the relevant standards and best practices of the international security literature for embedded systems are harmonized in one metamodel. By extending SAM, this research enables a more nuanced understanding and representation of vulnerabilities, attacks and scenarios, improving the effectiveness of security analysis and remediation strategies. The integration of different perspectives, including technological, social and behavioral factors, underlines the comprehensive nature of the proposed model. This interdisciplinary approach recognizes the complexity of security threats in autonomous vehicle systems and enables stakeholders to develop more effective countermeasures. This work presents an overview of the extension of the Security Abstration Model (SAM) and evaluates this security technology through expert interviews. The results demonstrate that our approach is compliant not only in the automotive domain, but also with standards for embedded systems from other domains. It thus provides a comprehensive and universal approach for the development of secure systems.
The proliferation of connected and autonomous vehicle technologies has significantly increased cybersecurity risks. Modern vehicles, as complex and networked computer systems, require comprehensive protection against malicious external attacks, much like conventional computers. Addressing these challenges requires robust tools that align established automotive model-based development approaches with the ISO/SAE 21434 standard for automotive cybersecurity, which became mandatory following its publication in 2021. Building on prior research, this paper introduces key innovations in the conceptual framework and the tool support that integrate seamlessly into existing automotive development methodologies. These advancements are rooted in extensions to the Security Abstraction Model (SAM) informed by the ISO/SAE 21434 standard. Notably, SAM now incorporates advanced methods for score calculation, including an attack potential-based approach for assessing attack feasibility and the computation of risk scores using risk matrices. Usability improvements are also a contribution, achieved through the introduction of BPMN-style (Business Process Model and Notation) diagrams tailored for the accessible visualization of otherwise complex security models. These diagrams make multifaceted attack trees easier to interpret, enabling managers and other non-technical stakeholders to intuitively understand security vulnerabilities and make informed decisions.
Additionally, the tool supports updated metrics for impact and risk analysis, demonstrated through practical applications involving automotive subsystems such as braking. These examples illustrate improved traceability between SAM and functional design, ensuring that cybersecurity requirements are effectively integrated into the broader development lifecycle.
We propose a meta-language-based approach enabling Large Language Models (LLMs) to reliably generate structured, machine-readable artifacts referred to as Meta-Language-defined Structures (MLDS) adapted to domain requirements, without adhering strictly to standard formats like JSON or XML. By embedding explicit schema instructions within prompts, we evaluated the method across diverse use cases, including automated Virtual Reality environment generation and automotive security modeling. Our experiments demonstrate that the meta-language approach significantly improves LLM-generated structure compliance, with an 88 % validation rate across 132 test scenarios. Compared to traditional methods using LangChain and Pydantic, our MLDS method reduces setup complexity by approximately 80 %, despite a marginally higher error rate. Furthermore, the MLDS artifacts produced were easily editable, enabling rapid iterative refinement. This flexibility greatly alleviates the “blank page syndrome” by providing structured initial artifacts suitable for immediate use or further human enhancement, making our approach highly practical for rapid prototyping and integration into complex industrial workflows.
Automatic speech recognition (ASR) for pathologic speech remains a major challenge due to high variability in articulation, phonation, and prosody distortions. In this work, we propose a pathology-aware speech encoder based on BEST-RQ pre-training, which incorporates 46k hours of speech, including
pathologic and atypical speech. We continue pre-training for domain adaptation and experiment with etiology-specific codebooks. We achieve a 13.2% relative word error rate (WER) improvement using the pathology-aware speech encoder with etiology-specific continued pre-training. Additionally, we examine the impact of incorporating synthetic and out-of-domain (OOD) data to further enhance ASR performance. Synthetic data reduces WER by up to 8.7%, while OOD data improves WER by 12.2%. Finally, we introduce a semantic similaritybased data augmentation technique to optimize data selection, achieving a WER improvement of up to 9.7% while minimizing
the need for additional training data.
We analyze different errors in speech recognition systems, focusing on consecutive insertions and deletions, known as hallucinations and elisions in
transformer-based end-to-end automatic speech recognition (ASR) systems. We compare errors from a TDNN-HMM, and whisper-based models on English and German spontaneous speech. Based on a human annotated subset of German lecture videos, we investigate whether these blocks of deletions affect the semantics of the utterance. Whisper performs best and preserves the meaning in 90% of the annotated error segments even containing consecutive deletions on this subset. We analyze the word error rate and do further analysis of errors using natural language processing to detect lemmatization errors, compound word errors, and out-of-vocabulary words. We discuss possible reasons and mitigations.