BEGIN:VCALENDAR
PRODID;X-RICAL-TZSOURCE=TZINFO:-//com.denhaven2/NONSGML ri_cal gem//EN
CALSCALE:GREGORIAN
VERSION:2.0
BEGIN:VEVENT
DTEND;VALUE=DATE-TIME:20221104T151500Z
DTSTART;VALUE=DATE-TIME:20221104T141500Z
DTSTAMP;VALUE=DATE-TIME:20250905T192738Z
UID:f9cc49fc-3bdc-4054-a12e-83bcdc79e031@talks.stuts.de
DESCRIPTION:An innocent letter-guessing-game called Wordle went viral in 
 the beginning of 2022. One\nparticular clone of this game\, called Seman
 tle1\, is particularly interesting from a linguistic\nperspective. The u
 ser is tasked with guessing a secret word based on its semantic similari
 ty to\nother words. Semantle is premised on The Distributional Hypothesi
 s (Firth\, 1957)\, which states\nthat the meaning of a lexical item can 
 be approximated by knowing the linguistic contexts in\nwhich it is used.
  Distributional Semantics concerns itself with building accurate distrib
 utional\nlexical representations from language corpora. A very intuitive
  and successful way of representing\nwords distributionally is using vec
 tor spaces (Clark\, 2015\; Turney & Pantel\, 2010). Today\, due\nto thei
 r attractive mathematical properties\, vector-based representations of l
 exical items (word\nembeddings) are an indispensable tool for virtually 
 all NLP tasks\, for example question answering\n(Karpukhin et al.\, 2020
 ) and co-reference resolution (Lee et al.\, 2017). In this workshop\, ai
 med at\nlinguists\, participants will be familiarized with the basic con
 cepts of count-based distributional\nmodels of lexical meaning (largely 
 following the structure of Turney and Pantel\, 2010) and\nwill become ac
 quainted with the more recent implementations of word-embeddings such as
 \nWord2Vec (Mikolov et al.\, 2013) and Fasttext (Bojanowski et al.\, 201
 7)\, as well as the basics\nof contextualized embeddings that are retrie
 ved from transformer models like GPT-3 (Peters\net al.\, 2018\; Radford 
 et al.\, 2019). In this workshop\, we will specifically look at and repl
 icate\nan intuitive and fun application of distributional semantics\, na
 mely Semantle. By trying to\nunderstand and implement what Semantle does
  and what makes it fun\, we will try to gain a\nsolid grasp of the repre
 sentations that power most NLP technologies in 2022.
URL:https://talks.stuts.de/de/stuts72/public/events/879
SUMMARY:Build your own Semantle-clone
ORGANIZER:stuts72
LOCATION:stuts72 - Wiwi-Bunker —Room 3035
END:VEVENT
END:VCALENDAR