BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Learning disentangled representation for interpretable language mo
 del. / Interactive Narrative Understanding. - Lin Gui and Runcong Zhao\, K
 ing's College London
DTSTART:20231019T100000Z
DTEND:20231019T110000Z
UID:TALK207397@talks.cam.ac.uk
CONTACT:Panagiotis Fytas
DESCRIPTION:Recent years have witnessed increasing interest in developing 
 interpretable models in Natural Language Processing (NLP). Most existing m
 odels aim at identifying input features such as words or phrases important
  for model predictions. Neural models developed in NLP\, however\, often c
 ompose word semantics in a disentangled manner. As such\, interpretation b
 y words or phrases only cannot faithfully explain model decisions. In our 
 recent work\, we propose a series of disentangle representation learning m
 ethods for interpretable language models\, including the interpretation in
  text classification with hierarchical explanation\, the uncertainty estim
 ation in prediction\, and controllable language generation with disentangl
 ement. Experimental results on real world datasets show that our proposed 
 approaches are able to generate interpretations more faithful to model pre
 dictions and better understood by humans.\n\nLarge language models (LLMs) 
 can be used to generate human-like responses\, offering a promising avenue
  for creating immersive and interactive environments. These environments h
 ave the potential to emulate the dynamic storylines readers might encounte
 r in books\, similar to those portrayed in the television series "Westworl
 d." However\, the capability of LLMs to truly grasp an author's intent rem
 ains a challenge. Narrative understanding seeks to capture the cognitive p
 rocesses of authors\, shedding light on their knowledge\, intentions\, bel
 iefs\, and desires. \n\nWe will introduce our recent work\, NarrativePlay\
 , a system that allows users to role-play a fictional character and intera
 ct with other characters in narratives such as novels in an immersive envi
 ronment\, guided by personality traits extracted from narratives. We also 
 incorporate automatically generated visual displays of narrative settings\
 , character portraits\, and character speech\, greatly enhancing the overa
 ll user experience.  
LOCATION:GR05\, English Faculty Building\, 9 West Road\, Sidgwick Site
END:VEVENT
END:VCALENDAR