Team members:
Katarina Aleksandra Brezovar
,18223286
,[email protected]
Klara Vrabl
,18223311
,[email protected]
Nives Hüll
,18223292
,[email protected]
Group public acronym/name: SWEAT
(Slovene Word sEnse disAmbiguaTion)
This value will be used for publishing marks/scores. It will be known only to you and not you colleagues.
In the folder gradnja-polisemnih
there is a program that generates the list polysemous_words that is the basis for finding appropriate candidates for word sense disambiguation.
In the folder iskanje-lem
there is another readme that explains in detail how to use the programs in the two folders and what is generated.
In the subfolder polisemne-leme-in-stavki
there is a program for finding centroids that served as the basis for choosing the appropriate candidates for the manually annotated WiC dataset.
It is evident from the first column of preformatiran-manual-evaluation.tsv
which lemma is being examined for WiC. If there are two word forms from the same lemma in the same sentence, the first one was taken into account.
In the folder reports
includes all the reports of our work.