TACRED Dataset


Learn more at https://nlp.stanford.edu/projects/tacred/
TACRED is a large-scale relation extraction dataset with ~106k samples built over newswire and online web text from the corpus that was initially used at the annual TAC KBP (Knowledge Base Population) challenges. The samples in TACRED cover 41 relation types (e.g., per:schools_attended and org:members) or are labeled as no_relation if there are no defined relations held. The samples were developed by combining available human annotations from the TAC KBP challenges with crowdsourcing data.


Download Instructions

You can download TACRED from the https://catalog.ldc.upenn.edu/LDC2018T24. If you are an LDC member, the access will be free; otherwise, an access fee of $25 is needed.