كانت تجربة مثيرة لعرض أول ملصق علمي خلال رحلتي للدكتوراة في مؤتمر افتراضي #syntaxfest2021
My first poster presentation during my PhD. in a virtual conference
It was so exciting!
There is still time to register to #SyntaxFest2021! We have an amazing program and it will all happen on our Gather space which includes a game room and a virtual Bulgaria space! More info on syntaxfest.github.io/syntaxf…
With the rapid spread of omicron, it seems clear that we unfortunately cannot hold #SyntaxFest2021 in person and it will therefore be held fully online, on zoom and gather.town. Details coming.
Super happy to share that our language paper (my first ever) for Old Turkish and Old Turkic will appear in #SyntaxFest2021@ud_workshop soon! This work covers lots of aspired elements for me, especially with the keyboard, transliteration scheme, and word definition for UPOS tags.
📰-alert: We looked at genre in the massively multilingual Universal Dependencies & examined to what degree genre can be predicted at the sentence level (@mxmeij, @robvanderg) #SyntaxFest2021arxiv.org/abs/2112.04971
It would be great to see more meta-data like genre in corpora
For #SyntaxFest2021 w/ @robvanderg and @barbara_plank, we ask:
How Universal is Genre in Universal Dependencies?
📄arxiv.org/abs/2112.04971
🧵Across 114 langs 18 genres, UD metadata are noisy, but a combination of local global features can recover the true genre distribution.
ALT In order to identify instance genre from treebank metadata, we cluster instances within treebanks first and compare UD-level genre using representations from mBERT.
For #SyntaxFest2021 w/ @robvanderg and @barbara_plank, we ask:
How Universal is Genre in Universal Dependencies?
📄arxiv.org/abs/2112.04971
🧵Across 114 langs 18 genres, UD metadata are noisy, but a combination of local global features can recover the true genre distribution.
ALT In order to identify instance genre from treebank metadata, we cluster instances within treebanks first and compare UD-level genre using representations from mBERT.
Preprint of our upcoming #TLT2021 @ #SyntaxFest2021 paper with @robvanderg! We evaluate dataset embeddings in multilingual dependency parsing models based on mBERT
#SyntaxFest2021: you should have received notifications now, contact us if not. Sorry again for the delay and congrats to all authors of accepted papers!
Dear CompLing and NLPers, the #SyntaxFest2021 is missing 10 reviews so if you'd like to provide some emergency reviews please let us know w/ @mdlhx@texttheater asap
DEADLINE EXTENSION - As hopefully many of you noticed, the deadline for #SyntaxFest2021 has been extended to Oct 11 and this includes UDW! We are very excited to read all your submissions! #nlproc
Can't believe I haven't seen tweets about the #SyntaxFest2021 :(
Guys, 4 conferences in a row: Quasi Workshop, UD workshop, Depling and TLT. Hybrid mode, Sophia Bulgaria. Deadline October 4th. Please share
syntaxfest.github.io/syntaxf…
Can't believe I haven't seen tweets about the #SyntaxFest2021 :(
Guys, 4 conferences in a row: Quasi Workshop, UD workshop, Depling and TLT. Hybrid mode, Sophia Bulgaria. Deadline October 4th. Please share
syntaxfest.github.io/syntaxf…