Multi-label text genre classification using Swedish corpora with pre-trained word vectors

Corpus

Using the COCTAILL corpus which contain aprox 800.000 tokens of Swedish texts from coursebooks aimed at second/foreign language (L2) learning. [1]

Word vectors

Model

References

[1] Volodina, Elena & Pilán, Ildikó & Eide, Stian & Heidarsson, Hannes. (2014). You get what you annotate: a pedagogically annotated corpus of coursebooks for Swedish as a Second Language.