BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//programme.europython.eu//europython-2024//speaker//CNDAR
 F
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-europython-2024-GHYHAE@programme.europython.eu
DTSTART;TZID=CET:20240710T104500
DTEND;TZID=CET:20240710T113000
DESCRIPTION:Customers only buy the products they are able to find. Improvin
 g the search functions on the website is crucial for user-friendliness.\n\
 nIn our talk we present the lessons learnt from improving the search of ou
 r global online marketplace\, which sells 20 million products per year. We
  moved from a traditional word-match based approach (BM25) to a modern hyb
 rid solution that combines BM25 with a semantic vector model\, an open-sou
 rce language model that we fine-tuned to our domain.\n\nWith numerous refe
 rences to current literature\, we will explain how we designed our new sys
 tem and solved the multiple challenges we encountered on both the ML and e
 ngineering side (data pipeline encoding documents\, live service encoding 
 queries\, integration with search engine). Our system is based on OpenSear
 ch\, the lessons can be applied to other search engines as well.\n\nIn par
 ticular the presentation will cover:\n- Status and Short-Comings of our ol
 d Search\n- Introduction of Hybrid Search\n- Our Machine Learning Solution
 \n- Architecture and Implementation (with special consideration of latency
 )\n- Learnings and Next Steps
DTSTAMP:20260513T164426Z
LOCATION:North Hall
SUMMARY:From Text to Context: How We Introduced a Modern Hybrid Search - An
 sgar Gruene\, Dharin Shah
URL:https://programme.europython.eu/europython-2024/talk/GHYHAE/
END:VEVENT
END:VCALENDAR