Highlighting the different functions it plays in text, this talk presents a versatile approach to detect and automatically annotate negation based on the two notions of trigger and its linguistic scope. The idea will be expanded to recognition of negation focus within its scope and extensions to speculative language and embedding constructions in general. The varied possibilities of interaction of different meta-factual constructions (negation, modality, attribution) will be shown to favor layered, general annotation approaches. A highly modular implementation has been tested on a variety of text types and evaluated in shared task competitions, in which it has been shown to port well across domains and text types.

Bio: Sabine Bergler is a Professor of Computer Science at Concordia University, where she directs the CLaC Lab (Computational Linguistics at Concordia). Her research has mainly focused on the meta-factual aspects of language, such as reported speech, negation, speculation, opinions, and sentiment and their semantic interpretation. Recent work focused on stand-alone, linguistically inspired modules, combined into progressively more complex application pipelines and extensively evaluated in shared task competitions. She holds a MS from University of Massachusetts at Amherst and a PhD in Computer Science from Brandeis University.