Neural Generation for Czech: Data and Baselines

TL;DR

This paper introduces a new Czech NLG dataset in the restaurant domain and proposes two baseline models to handle the language's morphological complexity, advancing non-English NLG research.

Contribution

It provides the first Czech NLG dataset and explores novel baseline models addressing morphological inflection challenges.

Findings

Baseline models achieve promising results on Czech NLG.

Morphological handling improves generation quality.

Two-step approach effectively manages inflection complexity.

Abstract

We present the first dataset targeted at end-to-end 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-of-the-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.