cs_restaurants / cs_restaurants.json
Sebastian Gehrmann
.
9eb854d
{
"overview": {
"where": {
"has-leaderboard": "no",
"leaderboard-url": "N/A",
"leaderboard-description": "N/A",
"website": "n/a",
"data-url": "[Github](https://github.com/UFAL-DSG/cs_restaurant_dataset)",
"paper-url": "[Github](https://aclanthology.org/W19-8670/)",
"paper-bibtext": "```\n@inproceedings{cs_restaurants,\n\taddress = {Tokyo, Japan},\n\ttitle = {Neural {Generation} for {Czech}: {Data} and {Baselines}},\n\tshorttitle = {Neural {Generation} for {Czech}},\n\turl = {https://www.aclweb.org/anthology/W19-8670/},\n\turldate = {2019-10-18},\n\tbooktitle = {Proceedings of the 12th {International} {Conference} on {Natural} {Language} {Generation} ({INLG} 2019)},\n\tauthor = {Du\u0161ek, Ond\u0159ej and Jur\u010d\u00ed\u010dek, Filip},\n\tmonth = oct,\n\tyear = {2019},\n\tpages = {563--574},\n}\n```",
"contact-name": "Ondrej Dusek",
"contact-email": "odusek@ufal.mff.cuni.cz"
},
"languages": {
"is-multilingual": "no",
"license": "cc-by-sa-4.0: Creative Commons Attribution Share Alike 4.0 International",
"task-other": "N/A",
"language-names": [
"Czech"
],
"intended-use": "The dataset was created to test neural NLG systems in Czech and their ability to deal with rich morphology.\n\n",
"language-speakers": "Six professional translators produced the outputs",
"language-dialects": "No breakdown of dialects is provided. ",
"license-other": "N/A",
"task": "Dialog Response Generation",
"communicative": "Producing a text expressing the given intent/dialogue act and all and only the attributes specified in the input meaning representation.\n\n"
},
"credit": {
"organization-type": [
"academic"
],
"organization-names": "Charles University, Prague",
"creators": "Ondrej Dusek and Filip Jurcicek",
"funding": "This research was supported by the Charles University project PRIMUS/19/SCI/10 and by the Ministry of Education, Youth and Sports of the Czech Republic under the grant agreement LK11221. This work used using language resources distributed by the LINDAT/CLARIN project of the Ministry of Education, Youth and Sports of the Czech Republic (project LM2015071).",
"gem-added-by": "Simon Mille wrote the initial data card and Yacine Jernite the data loader. Sebastian Gehrmann migrated the data card and loader to the v2 format. "
},
"structure": {
"data-fields": "The data is stored in a JSON or CSV format, with identical contents. The data has 4 fields:\n* `da`: the input meaning representation/dialogue act (MR)\n* `delex_da`: the input MR, delexicalized -- all slot values are replaced with placeholders, such as `X-name`\n* `text`: the corresponding target natural language text (reference)\n* `delex_text`: the target text, delexicalized (delexicalization is applied regardless of inflection)\n\nIn addition, the data contains a JSON file with all possible inflected forms for all slot values in the dataset (`surface_forms.json`).\nEach slot -> value entry contains a list of inflected forms for the given value, with the base form (lemma), the inflected form, and\na [morphological tag](https://ufal.mff.cuni.cz/pdt/Morphology_and_Tagging/Doc/hmptagqr.html).\n\nThe same MR is often repeated multiple times with different synonymous reference texts.\n",
"structure-splits": "| Property | Value |\n|--------------------------------|-------|\n| Total instances | 5,192 |\n| Unique MRs | 2,417 |\n| Unique delexicalized instances | 2,752 |\n| Unique delexicalized MRs | 248 |\n\nThe data is split in a roughly 3:1:1 proportion into training, development and test sections, making sure no delexicalized MR\nappears in two different parts. On the other hand, most DA types/intents are represented in all data parts.\n\n",
"structure-description": "The data originated as a translation and localization of [Wen et al.'s SF restaurant](https://www.aclweb.org/anthology/D15-1199/) NLG dataset.\n",
"structure-example": "```\n{\n \"input\": \"inform_only_match(food=Turkish,name='\u0160vejk Restaurant',near='Charles Bridge',price_range=cheap)\",\n \"target\": \"Na\u0161la jsem pouze jednu levnou restauraci pobl\u00ed\u017e Karlova mostu , kde pod\u00e1vaj\u00ed tureckou kuchyni , \u0160vejk Restaurant .\"\n}\n```",
"structure-labels": "The input MRs were collected from [Wen et al.'s SF restaurant](https://www.aclweb.org/anthology/D15-1199/) NLG data\nand localized by randomly replacing slot values (using a list of Prague restaurant names, neighborhoods etc.).\n\nThe generated slot values were then automatically replaced in reference texts in the data.\n",
"structure-splits-criteria": "The creators ensured that after delexicalization of the meaning representation there was no overlap between training and test. \n\nThe data is split at a 3:1:1 rate between training, validation, and test.",
"structure-outlier": "n/a"
},
"what": {
"dataset": "The Czech Restaurants dataset is a task oriented dialog dataset in which a model needs to verbalize a response that a service agent could provide which is specified through a series of dialog acts. The dataset originated as a translation of an English dataset to test the generation capabilities of an NLG system on a highly morphologically rich language like Czech. "
}
},
"curation": {
"original": {
"is-aggregated": "no",
"aggregated-sources": "N/A",
"rationale": "The dataset was created to test neural NLG systems in Czech and their ability to deal with rich morphology.\n\n",
"communicative": "Producing a text expressing the given intent/dialogue act and all and only the attributes specified in the input MR.\n\n"
},
"language": {
"found": [],
"crowdsourced": [],
"created": "Six professional translators translated the underlying dataset with the following instructions: \n\n- Each utterance should be translated by itself\n- fluent spoken-style Czech should be produced\n- Facts should be preserved\n- If possible, synonyms should be varied to create diverse utterances\n- Entity names should be inflected as necessary\n- the reader of the generated text should be addressed using formal form and self-references should use the female form.\n\nThe translators did not have access to the meaning representation.",
"machine-generated": "N/A",
"validated": "validated by data curator",
"is-filtered": "not filtered",
"filtered-criteria": "N/A",
"obtained": [
"Created for the dataset"
],
"producers-description": "n/a",
"topics": "n/a",
"pre-processed": "n/a"
},
"annotations": {
"origin": "none",
"rater-number": "N/A",
"rater-qualifications": "N/A",
"rater-training-num": "N/A",
"rater-test-num": "N/A",
"rater-annotation-service-bool": "no",
"rater-annotation-service": [],
"values": "N/A",
"quality-control": [],
"quality-control-details": "N/A"
},
"consent": {
"has-consent": "no",
"consent-policy": "N/A",
"consent-other": "N/A",
"no-consent-justification": "It was not explicitly stated but we can safely assume that the translators agreed to this use of their data. "
},
"pii": {
"has-pii": "no PII",
"no-pii-justification": "This dataset does not include any information about individuals.",
"is-pii-identified": "N/A",
"pii-identified-method": "N/A",
"is-pii-replaced": "N/A",
"pii-replaced-method": "N/A",
"pii-categories": []
},
"maintenance": {
"has-maintenance": "no",
"description": "N/A",
"contact": "N/A",
"contestation-mechanism": "N/A",
"contestation-link": "N/A",
"contestation-description": "N/A"
}
},
"gem": {
"rationale": {
"sole-task-dataset": "yes",
"sole-language-task-dataset": "yes",
"distinction-description": "The dialog acts in this dataset are much more varied than the e2e dataset which is the closest in style. ",
"contribution": "This is one of a few non-English data-to-text datasets, in a well-known domain, but covering a morphologically rich language that is harder to generate since named entities need to be inflected. This makes it harder to apply common techniques such as delexicalization or copy mechanisms.\n\n",
"model-ability": "surface realization"
},
"curation": {
"has-additional-curation": "yes",
"modification-types": [],
"modification-description": "N/A",
"has-additional-splits": "yes",
"additional-splits-description": "5 challenge sets for the Czech Restaurants dataset were added to the GEM evaluation suite.\n\n1. Data shift: We created subsets of the training and development sets of 500 randomly selected inputs each.\n2. Scrambling: We applied input scrambling on a subset of 500 randomly selected test instances; the order of the input dialogue acts was randomly reassigned.\n3. We identified different subsets of the test set that we could compare to each other so that we would have a better understanding of the results. There are currently two selections that we have made:\n\nThe first comparison is based on input size: the number of predicates differs between different inputs, ranging from 1 to 5.\nThe table below provides an indication of the distribution of inputs with a particular length.\nIt is clear from the table that this distribution is not balanced, and comparisions between items should be done with caution. \nParticularly for input size 4 and 5, there may not be enough data to draw reliable conclusions.\n\n| Input length | Number of inputs |\n|--------------|------------------|\n| 1 | 183 |\n| 2 | 267 |\n| 3 | 297 |\n| 4 | 86 |\n| 5 | 9 |\n\nThe second comparison is based on the type of act. Again we caution against comparing the different groups that have relatively few items.\nIt is probably OK to compare `inform` and `?request`, but the other acts are all low-frequent.\n\n| Act | Frequency |\n|-------------------|-----------|\n| ?request | 149 |\n| inform | 609 |\n| ?confirm | 22 |\n| inform_only_match | 16 |\n| inform_no_match | 34 |\n| ?select | 12 |\n\n\n\n\n",
"additional-splits-capacicites": "Generalization and robustness."
},
"starting": {
"research-pointers": "n/a",
"technical-terms": "- utterance: something a system or user may say in a turn\n- meaning representation: a representation of meaning that the system should be in accordance with. The specific type of MR in this dataset are dialog acts which describe what a dialog system should do, e.g., inform a user about a value. \n"
}
},
"results": {
"results": {
"other-metrics-definitions": "N/A",
"has-previous-results": "no",
"current-evaluation": "N/A",
"previous-results": "N/A",
"model-abilities": "Surface realization",
"metrics": [
"BLEU",
"ROUGE",
"METEOR"
],
"original-evaluation": "This dataset uses the suite of word-overlap-based automatic metrics from the E2E NLG Challenge (BLEU, NIST, ROUGE-L, METEOR, and CIDEr). In addition, the slot error rate is measured. "
}
},
"considerations": {
"pii": {
"risks-description": "n/a"
},
"licenses": {
"dataset-restrictions-other": "N/A",
"data-copyright-other": "N/A",
"dataset-restrictions": [
"open license - commercial use allowed"
],
"data-copyright": [
"open license - commercial use allowed"
]
},
"limitations": {
"data-technical-limitations": "The test set may lead users to over-estimate the performance of their NLG systems with respect to their generalisability, because there are no unseen restaurants or addresses in the test set. This is something we will look into for future editions of the GEM shared task.\n\n"
}
},
"context": {
"previous": {
"is-deployed": "no",
"described-risks": "N/A",
"changes-from-observation": "N/A"
},
"underserved": {
"helps-underserved": "yes",
"underserved-description": "The dataset may help improve NLG methods for morphologically rich languages beyond Czech.\n\n"
},
"biases": {
"has-biases": "yes",
"bias-analyses": "To ensure consistency of translation, the data always uses formal/polite address for the user, and uses the female form for first-person self-references (as if the dialogue agent producing the sentences was female). This prevents data sparsity and ensures consistent results for systems trained on the dataset, but does not represent all potential situations arising in Czech.\n\n"
}
}
}