BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.devconf.info//devconf-cz-2026//talk//F83XNQ
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-devconf-cz-2026-F83XNQ@pretalx.devconf.info
DTSTART;TZID=CET:20260618T121500
DTEND;TZID=CET:20260618T123000
DESCRIPTION:We will explore how to transform flat review data into a semant
 ically rich knowledge graph\, mapping entities and their multi-hop relatio
 nships. By leveraging KGNNs\, the system performs sophisticated relational
  retrieval that standard vector-only RAG misses. We’ll dive into:\n\nGra
 ph Construction: Building a scalable knowledge graph from Yelp's business 
 and review data.\n\nKGNN Integration: How Graph Neural Networks improve th
 e "Retrieval" phase by identifying deep-seated patterns and entity associa
 tions.\n\nThe RAG Pipeline: Connecting graph-based context to the LLM gene
 ration layer for more accurate\, fact-grounded recommendations.\n\nBenchma
 rking: Comparing traditional vector-RAG against KGNN-RAG in terms of reaso
 ning quality and factual consistency.\n\nAttendees will leave with a pract
 ical roadmap for implementing Graph-Augmented AI to solve complex\, multi-
 entity retrieval challenges in production.
DTSTAMP:20260430T125241Z
LOCATION:A113 (capacity 64)
SUMMARY:Bridging Structured Knowledge and LLMs: The Yelp-KGNN-RAG Architect
 ure - KEERTHI UDAYAKUMAR
URL:https://pretalx.devconf.info/devconf-cz-2026/talk/F83XNQ/
END:VEVENT
END:VCALENDAR