Word2vec cosine similarity
The popular word2vec algorithm (Raj et al., 2013) generates a vector (we use 300-dimensional vector representation) for each token in a corpus. The purpose of these vectors is usually to be used as features in downstream NLP tasks. But they can also be used for similarity. The original paper validates the vectors by testing them on word similarity tasks: the association score is the cosine between the vector for q and the vector for t. This score only applies to single-token q.