• Credible Open Knowledge Network (COKN)
• ClaimBuster: end-to-end fact-checking
  demo, API, Slackbot C+J16 talk video video 1 video 2 IJCNN18 KDD17 VLDB17 demo MetroCon17 talk

  First Runner-Up Award in the SIGMOD17 Undergraduate Student Research Competition (Damian Jimenez)

• ClaimPortal: integrated monitoring, searching, checking, and analytics of factual claims on twitter demo video ACL19 demo
• iCheck: fact-checking by perturbation analysis (led by Duke)
  ACM TODS VLDB14 SIGMOD14 demo

• FactWatcher: monitoring of facts from real-world events
  demo video VLDB14 demo

  VLDB14 Excellent demonstratione Award

  • Situational Fact Finding ICDE14 ICDE14 talk
  • Prominent Streak Discovery ACM TKDD KDD11
  • One-of-the-Few Objects KDD12
• Maverick: discovering exceptional facts from knowledge graphs
  demo SIGMOD18 talk video video SIGMOD18 SIGMOD18 talk VLDB18 demo
• Frequent episode mining (led by Chinese Academy of Sciences)
  ICDE15 WWW14 poster

Claims of "fact" are constantly made from data--by journalists, politicians, lobbyists, public relations specialists, sports fans, etc. Wherever numbers and data are involved, they can be laden with "lies, d--ed lies, and statistics." Database research has in the past focused on how to answer queries, but has not devoted much attention to discerning the quality of the resulting claims, or to formulating good queries from the outset. This project fills this void, by seeking to advance the understanding of what makes for a high-quality claim based on data, and how to find queries that lead there.

This project plans to develop a systematic solution for data-driven "fact-checking" (assessing quality of claims) and "lead-finding" (identifying claims or objects for further investigation), by addressing modeling, algorithmic, and systems challenges.

There is demand for this research in many domains where decisions are increasingly driven by data, e.g., public policy, business intelligence, homeland security, and health care. Nonetheless, this project chooses public interest journalism as its target domain, because it is one area where resources are severely strained and innovation is pressingly needed. In the past, traditional news organizations have provided public interest reporting to hold governments, corporations, and powerful individuals accountable. The decline of traditional media in recent years, however, has led to dwindling support for this vitally important type of journalism. Data-driven fact-checking and lead-finding are growing in importance, as more data become publicly available in the movement of "democratizing data." Taking advantage of data availability, this project hopes to reduce cost, increase effectiveness, and broaden participation for public interest journalism, by putting practical tools in the hands of journalists and citizens alike. Such tools help promote transparency in reporting, boost investigative activities for the underserved, and educate the public in data and quantitative analysis. In sum, this project represents a step towards "democratizing data analysis," complementing the current movement of democratizing data.

• Chengkai Li

• Fatma Dogan
• Naeemul Hassan
• Afroza Sultana
• Gensheng Zhang

• Minumol Joseph

• Bill Adair (Duke University)
• Pankaj Agarwal (Duke University)
• James T. Hamilton (Stanford University)
• Ping Luo (HP Labs, Chinese Academy of Sciences)
• Mark Tremayne (UT Arlington, Department of Communication)
• Min Wang (HP Labs, Google Research)
• Jun Yang (Duke University)
• Cong Yu (Google Research)