BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//pyconde-pydata-berlin-2023//speaker//8S77N
 V
BEGIN:VTIMEZONE
TZID:Europe/Berlin
BEGIN:DAYLIGHT
DTSTART:20220418T000000
TZNAME:CEST
TZOFFSETFROM:+0200
TZOFFSETTO:+0200
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20221030T030000
RDATE:20231029T030000
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20230326T030000
RDATE:20240331T030000
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
SUMMARY:Accelerating Public Consultations with Large Language Models: A Ca
 se Study from the UK Planning Inspectorate - Michele Dallachiesa\, Andreas
  Leed
DTSTART;TZID=Europe/Berlin:20230418T144500
DTEND;TZID=Europe/Berlin:20230418T153000
DTSTAMP:20260820T003449Z
UID:pretalx-pyconde-pydata-berlin-2023-HMGCPL@pretalx.com
DESCRIPTION:Local Planning Authorities (LPAs) in the UK rely on written re
 presentations from the community to inform their Local Plans which outline
  development needs for their area. With an average of 2000 representations
  per consultation and 4 rounds of consultation per Local Plan\, the volume
  of information can be overwhelming for both LPAs and the Planning Inspect
 orate tasked with examining the legality and soundness of plans. In this s
 tudy\, we investigate the potential for Large Language Models (LLMs) to st
 reamline representation analysis.\n\nWe find that LLMs have the potential 
 to significantly reduce the time and effort required to analyse representa
 tions\, with simulations on historical Local Plans projecting a reduction 
 in processing time by over 30%\, and experiments showing classification ac
 curacy of up to 90%. \n\nIn this presentation\, we discuss our experimenta
 l process which used a distributed experimentation environment with Jupyte
 r Lab and cloud resources to evaluate the performance of the BERT\, RoBERT
 a\, DistilBERT\, and XLNet models. We also discuss the design and prototyp
 ing of web applications to support the aided processing of representations
  using Voilà\, FastAPI\, and React. Finally\, we highlight successes and 
 challenges encountered and suggest areas for future improvement.
LOCATION:A1
URL:https://pretalx.com/pyconde-pydata-berlin-2023/talk/HMGCPL/
END:VEVENT
END:VCALENDAR