We present a new hybrid knowledge base that combines the contextual information of distributional models with the conciseness and precision of manually constructed lexical networks. In contrast to dense vector representations, our resource is human readable and interpretable, and can be easily embedded within the Semantic Web ecosystem.