Research on definition extraction has been conducted for well over a decade,\nlargely with significant constraints on the type of definitions considered. In\nthis work, we present DeftEval, a SemEval shared task in which participants\nmust extract definitions from free text using a term-definition pair corpus\nthat reflects the complex reality of definitions in natural language.\nDefinitions and glosses in free text often appear without explicit indicators,\nacross sentences boundaries, or in an otherwise complex linguistic manner.\nDeftEval involved 3 distinct subtasks: 1)Sentence classification, 2) sequence\nlabeling, and 3) relation extraction.\n