Let Guidelines Guide You: A Prescriptive Guideline-Centered Data Annotation Methodology
Introduces GCAM, a methodology that associates annotation guidelines with individual data instances.
federico-ruggeri
Introduces GCAM, a methodology that associates annotation guidelines with individual data instances.
Uses genetic-based learning to avoid interlocking in selective-rationalization pipelines.
Introduces RECV, a benchmark for evaluating deductive and abductive reasoning in evidence-based claim verification.
Analyzes how hate-speech definitions vary across cultures and domains.
Proposes responsible-development practices for speech datasets used in mental-health and neurological-disorder research.
Frames implicit misogyny detection as argumentative reasoning in Italian and English.
Introduces PejorativITy, an Italian resource for disambiguating pejorative epithets in misogyny detection.
Introduces annotation guidelines and an English news corpus for sentence-level subjectivity detection.
Investigates multimodal fallacy classification in political debates.
Studies disruptive-situation detection on public transport using speech emotion recognition.