ClaimBuster: A Benchmark Dataset of Check-worthy Factual Claims
The ClaimBuster dataset consists of statements extracted from all U.S. general election presidential debates (1960-2016) along with human-annotated check-worthiness labels. It contains 23,533 sentences where each sentence is categorized into one of the three categories: non-factual statement, unimportant factual statement, and check-worthy factual statement.
The work is partially supported by NSF grants IIS-1408928, IIP-1565699, IIS-1719054, OIA-1937143, a Knight Prototype Fund from the Knight Foundation, and subawards from Duke University as part of a grant to the Duke Tech & Check Cooperative from the Knight Foundation and Facebook. Any opinions, findings, and conclusions or recommendations expressed in this publication are those of the authors and do not necessarily reflect the views of the funding agencies.