1) A slot description encoder is employed to encode the slot descriptions into semantic embeddings, then (2) a BiLSTM is used to extract contextual neural features, and eventually (3) the outputs of both the slot description encoder and the BiLSTM are combined to outline a novel potential function in the CRF. Section 5.3.5 reveals that NSD performs higher when open vocabulary slots are handled as novel slots, and Section 5.3.