BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//euroscipy-2026//speaker//MR8NT3
BEGIN:VTIMEZONE
TZID:Poland
BEGIN:DAYLIGHT
DTSTART:20250720T000000
TZNAME:CEST
TZOFFSETFROM:+0200
TZOFFSETTO:+0200
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20251026T030000
RDATE:20261025T030000
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20260329T030000
RDATE:20270328T030000
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
SUMMARY:Finding the Right ROR: Semantic Search for Research Institutions -
  Diogo Rodrigues
DTSTART;TZID=Poland:20260720T152000
DTEND;TZID=Poland:20260720T155000
DTSTAMP:20260911T202636Z
UID:pretalx-euroscipy-2026-KFBJXK@pretalx.com
DESCRIPTION:Mapping freeform research affiliations to persistent identifie
 rs such as [ROR (Research Organization Registry)](https://ror.org/)  is ha
 rder than it looks. Institution names appear in many forms such as abbrevi
 ations\, alternate spellings\, local languages\, or legacy names\, thus ma
 king a reliable mapping difficult to achieve at scale.\n\nIn this talk\, w
 e present a semantic retrieval pipeline that reframes institution identifi
 cation as a search problem rather than a string-matching task. Our system 
 combines named entity recognition to extract institution entities\, dense 
 embeddings to represent their semantic meaning\, and vector search to retr
 ieve the most likely ROR matches. This approach allows us to handle noisy
 \, incomplete\, and multilingual inputs while remaining resilient to varia
 tion in how institutions are referenced.\n\nBy treating institution matchi
 ng as semantic retrieval\, we improve recall and robustness without relyin
 g on heuristics or on a continuous expanding rule-based approach. The syst
 em scales naturally as new institutions are added and as naming convention
 s evolve\, making it well suited for the dynamic research environment.\n\n
 We will share implementation details\, evaluation results\, and practical 
 lessons learned from deploying this pipeline in a real-world production se
 tting.
LOCATION:Room 1.19 (Ground Floor\, Shannon)
URL:https://pretalx.com/euroscipy-2026/talk/KFBJXK/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Boring AI Works: When BERT Beats Billion-Parameter Models - Diogo 
 Rodrigues
DTSTART;TZID=Poland:20260721T113000
DTEND;TZID=Poland:20260721T115000
DTSTAMP:20260911T202636Z
UID:pretalx-euroscipy-2026-RBUGDR@pretalx.com
DESCRIPTION:Recent advances in AI have shifted industries’ attention tow
 ard integrating LLM-based systems. Even though LLMs can solve a wide range
  of business problems\, they came with a significant complexity overhead. 
 At same time\, many real-world business applications involve well-defined 
 objectives\, predictable inputs\, and clear evaluation criteria. \n\nToday
 \, we are increasingly seeing a default pattern: for almost any NLP use ca
 se\, teams prompt GPT-like models and pay the bill at the end of the month
 . However\, this approach often introduces unnecessary complexity\, costs
 \, and operational risk. Many business and research problems exist in cons
 trained environments that can be solved with simpler techniques\, achievin
 g the same or higher success rates.\n\nThis talk defends that fine-tuned B
 ERT-based models remain a strong and often superior choice for targeted bu
 siness use cases that require NLP-based solutions. I propose to present a 
 real\, in-production use case where a simple transformer-based classifier 
 demonstrates a more favourable performance-cost trade-off than LLM-based a
 pproaches\, driven by lower latency\, reduced operational complexity\, eas
 ier fine-tuning\, and significantly lower maintenance costs.\n\nThe goal o
 f this presentation is not to reject LLMs\, but to promote a pragmatic\, o
 utcome-driven approach to NLP\, where “boring” solutions often deliver
  the most value.
LOCATION:Room 1.19 (Ground Floor\, Shannon)
URL:https://pretalx.com/euroscipy-2026/talk/RBUGDR/
END:VEVENT
END:VCALENDAR