A Statistical Model for Near-Synonym Choice - ResearchGate