Video Recordings

'Deep-syntactic' dependency structures that capture the argumentative, attributive and coordinative relations between full words of a sentence have a great potential for a number of NLP-applications. The abstraction degree of these structures is in between the output of a syntactic dependency parser (connected trees dened over all words of a sentence and language-specic grammatical functions) and the output of a semantic parser (forests of  trees dened over individual lexemes or phrasal chunks and abstract semantic role labels which capture the frame structures of predicative elements and drop all attributive and coordinative dependencies). In this talk, we will discuss some advantages of deep-syntactic  structures and present a parsing approach that provides deep syntactic structures.