Annotated Corpus of Polish Spoken Dialogues

摘要

The paper presents a corpus of Polish spoken dialogues being a result of the LUNA (spoken Language UNderstanding in multilinguAl communication systems) project. We describe the process of collecting the corpus and its annotation on several levels, from transcription of dialogues and their morphosyntactic analysis, to semantic annotation on concepts and predicates. Annotation on the morphosyntactic and se mantic levels was done automatically and then manually corrected. At the concept level, the annotation scheme comprises about 200 concepts from an ontology designed specially for the project. The set of frames for predicate level annotation was defined as a FrameNet-like resource.