Neural Generation for Czech: Data and Baselines
Ondřej Dušek, Filip Jurčíček · 2019
We present the first dataset targeted at end-toend NLG in Czech in the restaurant domain, along with several strong baseline models using the sequence-to-sequence approach.While non-English NLG is under-explored in general, Czech, as a morphologically rich language, makes the task even harder: Since Czech requires inflecting named entities, delexicalization or copy mechanisms do not work out-ofthe-box and lexicalizing the generated outputs is non-trivial.In our experiments, we present two different approaches to this this problem: (1) using a neural language model to select the correct inflected form while lexicalizing, (2) a two-step generation setup: our sequence-to-sequence model generates an interleaved sequence of lemmas and morphological tags, which are then inflected by a morphological generator. Hledáte vhodnou restauraci na X-good_for_meal ? Do-you-look-for a-suitable restaurant for [breakfast]X-name najdete v oblasti X-area .[Baráčnická rychta] you-find in the-area [of-Malá Strana] Chcete najít restauraci, kde se dobře X-good_for_meal ?Do-you-want to-find a-restaurant where yourself well [you-will-have-breakfast]X-name je na X-area . [Baráčnická rychta] is in [Malá Strana]Malá Strana NNFS1-----A---- Malé Strany