Unsupervised Metaphor Identification Using Hierarchical Graph Factorization Clustering

We present a novel approach to automatic metaphor identification, that discovers both metaphorical associations and metaphorical expressions in unrestricted text. Our system first performs hierarchical graph factorization clustering (HGFC) of nouns and then searches the resulting graph for metaphorical connections between concepts. It then makes use of the salient features of the metaphorically connected clusters to identify the actual metaphorical expressions. In contrast to previous work, our method is fully unsupervised. Despite this fact, it operates with an encouraging precision (0.69) and recall (0.61). Our approach is also the first one in NLP to exploit the cognitive findings on the differences in organisation of abstract and concrete concepts in the human brain.