Explicit world-knowledge and distributional semantic representations

This is an interdisciplinary course intended to bring together students from psycholinguistic and computational backgrounds and explore the question of world-knowledge in distributional semantics through lectures on recent published research. Distributional semantics exploits co-occurrences in corpus data in order to represent semantic knowledge implicitly through statistics about word context, but the extent to which this can serve as a proxy for semantic grounding in some form of world-knowledge is still an unresolved question. What we currently understand and how to think about the boundary between distributionally-represented knowledge and explicit world-knowledge will be the main topic of the course.