Smoothing dense spaces for improved relation extraction between drugs and adverse reactions.

•The goal is to extract Adverse Drug Reactions (ADRs) from Electronic Health Records (EHRs) in Spanish.•An ADR is a relation between two entities (a drug and, typically, a disease).•The challenge rests on the lexical variability in documents written at consultation time and on data insufficiency on the EHR domain.•Our contribution is to explore and assess alternative representations on continuous spaces to tackle lexical variability and simple smoothing to overcome data insufficiency.