PERLEX: A Bilingual Persian-English Gold Dataset for Relation Extraction

Table 3

PERLEX dataset statistics.

Dataset	Partition	Sentences	Words	Entities	Entity words	Average sentence length
PERLEX	Train	8,000	161,001	16,000	19,863	20.13
PERLEX	Test	2,717	54,633	5,434	6,699	20.11
PERLEX	Total	10,717	215,634	21,434	27,210	20.12
SemEval2010T8	Train	8,000	150,376	16,000	16,728	18.80
SemEval2010T8	Test	2,717	51,190	5,434	5,655	18.84
SemEval2010T8	Total	10,717	201,566	21,434	27,376	18.81