BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Knowledge Issues and Language Models  - James Thorne (KAIST)
DTSTART:20230616T110000Z
DTEND:20230616T120000Z
UID:TALK200041@talks.cam.ac.uk
CONTACT:Michael Schlichtkrull
DESCRIPTION:Abstract:\n\nAs we continue to push the boundaries of natural 
 language processing\, it becomes imperative to better understand how langu
 age models interact with\, incorporate\, and are influenced by knowledge. 
 In this talk\, I will navigate the complex interaction between language mo
 dels and knowledge.\n\nI'll begin by presenting a forthcoming ACL paper\, 
 studying fact verification against knowledge graphs. This demonstrates the
  ability to ground language model outputs against structured information s
 ources. Following this\, I'll discuss ongoing research in multi-hop multi-
 set retrieval settings\, and then explore how the abundance of knowledge a
 vailable can potentially undermine retrieve-and-reason architectures. Last
 ly\, I'll put forward a hypothesis about the role of knowledge in language
  models and consider the potential advantages of combining generative and 
 retrieval-based NLP approaches.\n\nBio:

\n\nJames is Assistant Profes
 sor at the KAIST Graduate School of AI\, South Korea\, working on large-sc
 ale and knowledge-intensive natural language understanding. James recently
  completed his PhD at the University of Cambridge where he developed model
 s and methods for automated fact verification and correction.
LOCATION:Computer Lab\, SS03
END:VEVENT
END:VCALENDAR