Thu. Sep 19th, 2024

Researchers from the University of California, Santa Barbara, and Tencent have released a dataset of 118,000 statements and 17,000 related Wikipedia tables to spur the development of systems that can verify facts using semi-structured data as evidence. The statements are labeled as entailed or refuted based on their veracity according to the tables. For example, a statement may say that John Smith was re-elected to Congress while a table lists the re-elected congressional members.

Get the data.

Forward to your friends