Refine
Year of publication
- 2023 (237) (remove)
Document Type
- Bachelor Thesis (93)
- Article (68)
- Conference Proceeding (30)
- Part of a Book (22)
- Book (10)
- Master's Thesis (2)
- Patent (2)
- Preprint (2)
- Talk (2)
- Contribution to a Periodical (1)
- Habilitation (1)
- Other (1)
- Part of Periodical (1)
- Report (1)
- Administrative publication (1)
Keywords
- Corporate Design (6)
- Editorial (6)
- Illustration (6)
- Typografie (6)
- Nachhaltigkeit (5)
- Produktdesign (5)
- Fotografie (4)
- Publikation (4)
- Design (3)
- Erscheinungsbild (3)
- Information extraction (3)
- Kinder (3)
- Künstliche Intelligenz (3)
- Museum (3)
- Natural language processing (3)
- nachhaltig (3)
- Animation (2)
- Architektur (2)
- Associated liquids (2)
- Bacillaceae (2)
Institute
- Fachbereich Gestaltung (95)
- Fachbereich Medizintechnik und Technomathematik (38)
- Fachbereich Elektrotechnik und Informationstechnik (23)
- ECSM European Center for Sustainable Mobility (21)
- Fachbereich Luft- und Raumfahrttechnik (20)
- Fachbereich Wirtschaftswissenschaften (17)
- Fachbereich Chemie und Biotechnologie (14)
- Fachbereich Energietechnik (14)
- INB - Institut für Nano- und Biotechnologien (12)
- Fachbereich Maschinenbau und Mechatronik (10)
- IfB - Institut für Bioengineering (9)
- Nowum-Energy (7)
- MASKOR Institut für Mobile Autonome Systeme und Kognitive Robotik (6)
- Fachbereich Bauingenieurwesen (5)
- Kommission für Forschung und Entwicklung (3)
- Solar-Institut Jülich (3)
- FH Aachen (2)
- Fachbereich Architektur (2)
- Institut fuer Angewandte Polymerchemie (2)
- Arbeitsstelle fuer Hochschuldidaktik und Studienberatung (1)
In recent years, the development of large pretrained language models, such as BERT and GPT, significantly improved information extraction systems on various tasks, including relation classification. State-of-the-art systems are highly accurate on scientific benchmarks. A lack of explainability is currently a complicating factor in many real-world applications. Comprehensible systems are necessary to prevent biased, counterintuitive, or harmful decisions.
We introduce semantic extents, a concept to analyze decision patterns for the relation classification task. Semantic extents are the most influential parts of texts concerning classification decisions. Our definition allows similar procedures to determine semantic extents for humans and models. We provide an annotation tool and a software framework to determine semantic extents for humans and models conveniently and reproducibly. Comparing both reveals that models tend to learn shortcut patterns from data. These patterns are hard to detect with current interpretability methods, such as input reductions. Our approach can help detect and eliminate spurious decision patterns during model development. Semantic extents can increase the reliability and security of natural language processing systems. Semantic extents are an essential step in enabling applications in critical areas like healthcare or finance. Moreover, our work opens new research directions for developing methods to explain deep learning models.