- Описание:
Чешский набор данных для преобразования текста в текст в ресторанном домене. Представления значений ввода содержат тип диалогового действия (сообщить, подтвердить и т. Д.), Слоты (еда, площадь и т. Д.) И их значения. Он возник как перевод набора данных английских ресторанов Сан-Франциско Веном и др. (2015).
Домашняя страница: https://github.com/UFAL-DSG/cs_restaurant_dataset
Исходный код:
tfds.structured.cs_restaurants.CSRestaurants
Версии:
-
1.0.0
( по умолчанию): Нет Замечания к выпуску.
-
Скачать Размер:
1.40 MiB
Dataset Размер:
2.46 MiB
Авто-кэшируются ( документация ): Да
расколы:
Расколоть | Примеры |
---|---|
'test' | 842 |
'train' | 3,569 |
'validation' | 781 |
- Особенности:
FeaturesDict({
'delex_input_text': FeaturesDict({
'table': Sequence({
'column_header': tf.string,
'content': tf.string,
'row_number': tf.int16,
}),
}),
'delex_target_text': tf.string,
'input_text': FeaturesDict({
'table': Sequence({
'column_header': tf.string,
'content': tf.string,
'row_number': tf.int16,
}),
}),
'target_text': tf.string,
})
Контролируемые ключи (см
as_supervised
документ ):('input_text', 'target_text')
Рис ( tfds.show_examples ): Не поддерживается.
Примеры ( tfds.as_dataframe ):
- Образец цитирования:
@inproceedings{dusek_neural_2019,
author = {Dušek, Ondřej and Jurčíček, Filip},
title = {Neural {Generation} for {Czech}: {Data} and {Baselines} },
shorttitle = {Neural {Generation} for {Czech} },
url = {https://www.aclweb.org/anthology/W19-8670/},
urldate = {2019-10-18},
booktitle = {Proceedings of the 12th {International} {Conference} on {Natural} {Language} {Generation} ({INLG} 2019)},
month = oct,
address = {Tokyo, Japan},
year = {2019},
pages = {563--574},
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.},
}