Word Embedding Calculus in Meaningful Ultradense Subspaces

Sascha Rothe, Hinrich Schütze
2016 Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)  
We decompose a standard embedding space into interpretable orthogonal subspaces and a "remainder" subspace. We consider four interpretable subspaces in this paper: polarity, concreteness, frequency and part-of-speech (POS) subspaces. We introduce a new calculus for subspaces that supports operations like "−1 × hate = love" and "give me a neutral word for greasy" (i.e., oleaginous). This calculus extends analogy computations like "king−man+woman = queen". For the tasks of Antonym Classification
more » ... nd POS Tagging our method outperforms the state of the art. We create test sets for Morphological Analogies and for the new task of Polarity Spectrum Creation.
doi:10.18653/v1/p16-2083 dblp:conf/acl/RotheS16 fatcat:ewxoguoz3fcoxfqxjkjnm6hyyq