BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Claim-Dissector: An Interpretable Fact-Checking System with Joint 
 Re-ranking and Veracity Prediction - Martin Fajčík ( Brno University of 
 Technology )
DTSTART:20220712T130000Z
DTEND:20220712T140000Z
UID:TALK176480@talks.cam.ac.uk
CONTACT:Michael Schlichtkrull
DESCRIPTION:Abstract:\n\nWe present Claim-Dissector: a novel latent variab
 le model for fact-checking and fact-analysis\, which given a claim and a s
 et of retrieved provenances allows learning jointly (i) what are the prove
 nances relevant to this claim (ii) what is the veracity of this claim. We 
 show that our system achieves state-of-the-art results on FEVER comparable
  to two-stage systems often used in traditional fact-checking pipelines\, 
 while using significantly less parameters and computation.\nOur analysis s
 hows that proposed approach further allows to learn not just which provena
 nces are relevant\, but also which provenances lead to supporting and whic
 h toward denying the claim\, without direct supervision. This not only add
 s interpretability\, but also allows to detect claims with conflicting evi
 dence automatically. Furthermore\, we study whether our model can learn fi
 ne-grained relevance cues while using coarse-grained supervision. We show 
 that our model can achieve competitive sentence-recall while using only pa
 ragraph-level relevance supervision. Finally\, traversing towards the fine
 st granularity of relevance\, we show that our framework is capable of ach
 ieving strong token-level interpretability. To do this\, we present a new 
 benchmark focusing on token-level interpretability ―  humans annotate to
 kens in relevant provenances they considered essential when making their j
 udgement. Then we measure how similar are these annotations to tokens our 
 model is focusing on. Our code\, dataset and demo will be released online.
 \n\n\nBio:\n\nMartin Fajčík (read as Fay-Cheek) is a PhD candidate in Na
 tural Language Processing from Knowledge Technology Research Group active 
 at FIT-BUT in Brno\, Czech Republic\, advised by prof. Pavel Smrž (ž is 
 read like j in french "Jean"). From 2021\, he also works as a research ass
 istant in IDIAP research institute based in Martigny\, Switzerland. His Ph
 D work is focusing on open-domain knowledge processing\, mainly in questio
 n answering and fact-checking. He enjoys a good hikes and an informal disc
 ussions over tea.
LOCATION:Computer Lab\, FW26
END:VEVENT
END:VCALENDAR