Conference Proceeding
Refine
Year of publication
- 2023 (30) (remove)
Document Type
- Conference Proceeding (30) (remove)
Keywords
- Natural language processing (3)
- Associated liquids (2)
- Future Skills (2)
- Information extraction (2)
- Interdisciplinarity (2)
- Power plants (2)
- Sustainability (2)
- Active learning (1)
- Agent-based simulation (1)
- Agile development (1)
- Android (1)
- Anomaly detection (1)
- Automation (1)
- Building Automation (1)
- Business Process Intelligence (1)
- CO2 (1)
- Carbon Dioxide (1)
- Clustering (1)
- Connected Automated Vehicle (1)
- Control (1)
- Datasets (1)
- Decision theory (1)
- Deep learning (1)
- Design Thinking (1)
- Digital transformation (1)
- Digital triage (1)
- Digital twin (1)
- District data model (1)
- District energy planning platform (1)
- Education (1)
- Electrocardiography (1)
- Electrochemistry (1)
- Elicit (1)
- Energy Disaggregation (1)
- Energy market design (1)
- Energy storage (1)
- Energy system planning (1)
- Future skills (1)
- Home Assistant (1)
- Home Automation Platform (1)
- Instagram store (1)
- Interculturality (1)
- Key competences (1)
- Machine Learning (1)
- Market modeling (1)
- Micromix (1)
- Minimum Risk Manoeuvre (1)
- Mpc (1)
- Natural Language Processing (1)
- Natural language understanding (1)
- Navigation (1)
- Neural networks (1)
- Open Source (1)
- Operational Design Domain (1)
- PLS (1)
- Path-following (1)
- Performance (1)
- Personality (1)
- Process Model Extraction (1)
- Process optimization (1)
- Profile extraction (1)
- Prototype (1)
- Quality control (1)
- Query learning (1)
- Relation classification (1)
- Renewable energy integration (1)
- Reproducible research (1)
- Sensors comparison (1)
- Smart Building (1)
- Social impact measurement (1)
- Society (1)
- Software (1)
- Stress testing (1)
- Sustainable engineering education (1)
- Teamwork (1)
- Text Mining (1)
- Text mining (1)
- Time-series synchronization (1)
- Transdisciplinarity (1)
- Transformative Competencies (1)
- Transiton of Control (1)
- Triage-app (1)
- Trustworthy artificial intelligence (1)
- V2X (1)
- Wearable electronic device (1)
- active learning (1)
- aircraft engine (1)
- combustion (1)
- emission index (1)
- hydrogen (1)
- lab work (1)
- nitric oxides (1)
- professional skills (1)
- purchase factor (1)
- shopping behavior (1)
- structural equation model (1)
Institute
- Fachbereich Elektrotechnik und Informationstechnik (15)
- ECSM European Center for Sustainable Mobility (7)
- Fachbereich Energietechnik (5)
- Fachbereich Luft- und Raumfahrttechnik (4)
- Fachbereich Medizintechnik und Technomathematik (4)
- MASKOR Institut für Mobile Autonome Systeme und Kognitive Robotik (4)
- Fachbereich Wirtschaftswissenschaften (3)
- Kommission für Forschung und Entwicklung (3)
- Solar-Institut Jülich (3)
- Nowum-Energy (2)
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.