Evaluates gluten-free word embeddings of Hungarian in the word analogy task

What are gluten-free word embeddings? The idea of langue modeling with sub-word level units is not new either universally (for a summary, see Nemeskey 2017) or in Hungarian NLP (from phrase-based statistical machine translation to static word embeddings (Siklósi and Novák 2016 in Hungarian, Siklósi 2016 paywalled)). The pun gluten-free comes from the term agglutinative for languages that glue suffixes to word stems to mark grammatical features.

jelmondatával → jelmondat <poss> <cas<ins>>
akartak → akar <past> <plur>

Example by Nemeskey (2017).

This modest project evaluates gluten-free word embeddings in the Hungarian equivalent of the word analogy task (Mikolov+ 2013) for the purposes of the PhD thesis of Márton Makrai. The thesis introduces the task in Section 4.2.7., and the results of this project are reported in Section 7.4.