Neural Generation for Czech: Data and Baselines
By Ondrej Dusek, Filip Jurvc'ivcek. 2019
22 citations recorded by the source corpus.
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.
Subfields: Natural Language Processing
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Natural Language Processing
- Sebastian Gehrmann
- Thibault Sellam
- Percy Liang
- Mohit Bansal
- Ankur P. Parikh
- Elizabeth Clark
- Mihir Kale
- Mirella Lapata
Citations in the corpus
- The GEM Benchmark: Natural Language Generation, its Evaluation and Metrics (2021, background): “He also supplied baseline outputs for E2E, Czech Restaurants, and WebNLG.”
- Automatic Construction of Evaluation Suites for Natural Language Generation Datasets (2021, background): “Data-to-text CommonGen CG en Expand concept words into natural language [32] Czech Restaurant CR cs Verbalize an agents response from dialog act [33] E2E E2 en Describe a restaurant from key-value attributes [34] ToTTo TO en Describe highlighted cells in a table [20] WebNLG WNen/ru en,ru Verbalize s”
- Template Guided Text Generation for Task Oriented Dialogue (2020, methodology): “However, the low human scores on informativeness suggest that getting perfect scores on metrics like SER may not be a reliable way to judge factual accuracy.”
- CINS: Comprehensive Instruction for Few-shot Learning in Task-oriented Dialog Systems (2021, methodology): “2002) and Slot Error Rate (SER Dušek and Jurcicek (2019)) as metrics.”
- TaTa: A Multilingual Table-to-Text Dataset for African Languages (2022, background): “Multilingual Generation At present, there exist only few existing data-to-text datasets that cover languages beyond English (Gardent et al., 2017b; Kanerva et al., 2019; Dušek and Jurčíček, 2019).”