Unstructured or semistructured data sources that are useful for ADR discovery include medical literature, plain text parts of various medical records and reports, and discussions on adverse drug effects in social media. The processing of this type of data often involves data annotation and semantic mapping to well-defined concepts and terms.