MAPS-KB is a million-scale probabilistic simile knowledge base, covering 4.3 million triplets over 0.4 million terms from 70 GB corpora. It is designed for the tasks of simile detection and component extraction.
Source: