A Distributional Multi-word Thesaurus in Sketch Engine
Miloš Jakubíček, Pavel Rychlý · Zenodo (CERN European Organization for Nuclear Research) · 2019
In this paper we present an extension of the current distributional thesaurus as available in the Sketch Engine corpus management system towards multi-word units. We explain how multi-word sketches are used to generate multi-word unit candidates, thus preserving access to the underlying corpus texts. Finally we present sample results on the British National Corpus and discuss future development as well as difficulties in evaluation.