{ "overview": { "where": { "has-leaderboard": "no", "leaderboard-url": "N/A", "leaderboard-description": "N/A", "website": "[Website](https://sites.google.com/view/wngt19/dgt-task)", "data-url": "[Github](https://github.com/neulab/dgt)", "paper-url": "[ACL Anthology](https://www.aclweb.org/anthology/D19-5601/)", "paper-bibtext": "```\n@inproceedings{hayashi-etal-2019-findings,\n title = \"Findings of the Third Workshop on Neural Generation and Translation\",\n author = \"Hayashi, Hiroaki and\n Oda, Yusuke and\n Birch, Alexandra and\n Konstas, Ioannis and\n Finch, Andrew and\n Luong, Minh-Thang and\n Neubig, Graham and\n Sudoh, Katsuhito\",\n booktitle = \"Proceedings of the 3rd Workshop on Neural Generation and Translation\",\n month = nov,\n year = \"2019\",\n address = \"Hong Kong\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/D19-5601\",\n doi = \"10.18653/v1/D19-5601\",\n pages = \"1--14\",\n abstract = \"This document describes the findings of the Third Workshop on Neural Generation and Translation, held in concert with the annual conference of the Empirical Methods in Natural Language Processing (EMNLP 2019). First, we summarize the research trends of papers presented in the proceedings. Second, we describe the results of the two shared tasks 1) efficient neural machine translation (NMT) where participants were tasked with creating NMT systems that are both accurate and efficient, and 2) document generation and translation (DGT) where participants were tasked with developing systems that generate summaries from structured data, potentially with assistance from text in another language.\",\n}\n```", "contact-name": "Hiroaki Hayashi", "contact-email": "hiroakih@andrew.cmu.edu" }, "languages": { "is-multilingual": "yes", "license": "cc-by-4.0: Creative Commons Attribution 4.0 International", "task-other": "N/A", "language-names": [ "English", "German" ], "intended-use": "Foster the research on document-level generation technology and contrast the methods for different types of inputs.", "license-other": "N/A", "task": "Data-to-Text", "communicative": "Describe a basketball game given its box score table (and possibly a summary in a foreign language)." }, "credit": { "organization-type": [ "academic" ], "organization-names": "Carnegie Mellon University", "creators": "Graham Neubig (Carnegie Mellon University), Hiroaki Hayashi (Carnegie Mellon University)", "funding": "Graham Neubig", "gem-added-by": "Hiroaki Hayashi (Carnegie Mellon University)" }, "structure": { "data-fields": "- `id` (`string`): The identifier from the original dataset.\n- `gem_id` (`string`): The identifier from GEMv2.\n- `day` (`string`): Date of the game (Format: `MM_DD_YY`)\n- `home_name` (`string`): Home team name.\n- `home_city` (`string`): Home team city name.\n- `vis_name` (`string`): Visiting (Away) team name.\n- `vis_city` (`string`): Visiting team (Away) city name.\n- `home_line` (`Dict[str, str]`): Home team statistics (e.g., team free throw percentage).\n- `vis_line` (`Dict[str, str]`): Visiting team statistics (e.g., team free throw percentage).\n- `box_score` (`Dict[str, Dict[str, str]]`): Box score table. (Stat_name to [player ID to stat_value].)\n- `summary_en` (`List[string]`): Tokenized target summary in English.\n- `sentence_end_index_en` (`List[int]`): Sentence end indices for `summary_en`.\n- `summary_de` (`List[string]`): Tokenized target summary in German.\n- `sentence_end_index_de` (`List[int]`): ): Sentence end indices for `summary_de`.\n- (Unused) `detok_summary_org` (`string`): Original summary provided by RotoWire dataset.\n- (Unused) `summary` (`List[string]`): Tokenized summary of `detok_summary_org`.\n- (Unused) `detok_summary` (`string`): Detokenized (with organizer's detokenizer) summary of `summary`.\n", "structure-description": "- Structured data are directly imported from the original RotoWire dataset.\n- Textual data (English, German) are associated to each sample.", "structure-labels": "N/A", "structure-splits": "- Train\n- Validation\n- Test", "structure-example": "```\n{\n 'id': '11_02_16-Jazz-Mavericks-TheUtahJazzdefeatedthe',\n 'gem_id': 'GEM-RotoWire_English-German-train-0'\n 'day': '11_02_16',\n 'home_city': 'Utah',\n 'home_name': 'Jazz',\n 'vis_city': 'Dallas',\n 'vis_name': 'Mavericks',\n 'home_line': {\n 'TEAM-FT_PCT': '58', ...\n },\n 'vis_line': {\n 'TEAM-FT_PCT': '80', ...\n },\n 'box_score': {\n 'PLAYER_NAME': {\n '0': 'Harrison Barnes', ...\n }, ...\n 'summary_en': ['The', 'Utah', 'Jazz', 'defeated', 'the', 'Dallas', 'Mavericks', ...],\n 'sentence_end_index_en': [16, 52, 100, 137, 177, 215, 241, 256, 288],\n 'summary_de': ['Die', 'Utah', 'Jazz', 'besiegten', 'am', 'Mittwoch', 'in', 'der', ...],\n 'sentence_end_index_de': [19, 57, 107, 134, 170, 203, 229, 239, 266],\n 'detok_summary_org': \"The Utah Jazz defeated the Dallas Mavericks 97 - 81 ...\",\n 'detok_summary': \"The Utah Jazz defeated the Dallas Mavericks 97-81 ...\",\n 'summary': ['The', 'Utah', 'Jazz', 'defeated', 'the', 'Dallas', 'Mavericks', ...],\n}\n```", "structure-splits-criteria": "- English summaries are provided sentence-by-sentence to professional German translators with basketball knowledge to obtain sentence-level German translations.\n- Split criteria follows the original RotoWire dataset.", "structure-outlier": "- The (English) summary length in the training set varies from 145 to 650 words, with an average of 323 words." }, "what": { "dataset": "This dataset is a data-to-text dataset in the basketball domain. The input are tables in a fixed format with statistics about a game (in English) and the target is a German translation of the originally English description. The translations were done by professional translators with basketball experience. The dataset can be used to evaluate the cross-lingual data-to-text capabilities of a model with complex inputs. " } }, "curation": { "original": { "is-aggregated": "yes", "aggregated-sources": "RotoWire", "rationale": "A random subset of RotoWire dataset was chosen for German translation annotation.", "communicative": "Foster the research on document-level generation technology and contrast the methods for different types of inputs." }, "language": { "found": [], "crowdsourced": [], "created": "Professional German language translators were hired to translate basketball summaries from a subset of RotoWire dataset.", "machine-generated": "N/A", "validated": "validated by data curator", "is-filtered": "not filtered", "filtered-criteria": "N/A", "obtained": [ "Created for the dataset" ], "producers-description": "Translators are familiar with basketball terminology.", "topics": "Basketball (NBA) game summaries.", "pre-processed": "Sentence-level translations were aligned back to the original English summary sentences." }, "annotations": { "origin": "automatically created", "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": "Sentence-end indices for the tokenized summaries. Sentence boundaries can help users accurately identify aligned sentences in both languages, as well as allowing an accurate evaluation that involves sentence boundaries (ROUGE-L).", "quality-control": "validated through automated script", "quality-control-details": "Token and number overlaps between pairs of aligned sentences are measured." }, "consent": { "has-consent": "no", "consent-policy": "N/A", "consent-other": "N/A", "no-consent-justification": "Reusing by citing the original papers:\n- Sam Wiseman, Stuart M. Shieber, Alexander M. Rush:\nChallenges in Data-to-Document Generation. EMNLP 2017.\n- Hiroaki Hayashi, Yusuke Oda, Alexandra Birch, Ioannis Konstas, Andrew Finch, Minh-Thang Luong, Graham Neubig, Katsuhito Sudoh. Findings of the Third Workshop on Neural Generation and Translation. WNGT 2019." }, "pii": { "has-pii": "unlikely", "no-pii-justification": "N/A", "is-pii-identified": "no identification", "pii-identified-method": "N/A", "is-pii-replaced": "N/A", "pii-replaced-method": "N/A", "pii-categories": [ "generic PII" ] }, "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", "distinction-description": "The potential use of two modalities (data, foreign text) as input.", "sole-language-task-dataset": "yes", "contribution": "The use of two modalities (data, foreign text) to generate a document-level text summary.", "model-ability": "- Translation\n- Data-to-text verbalization\n- Aggregation of the two above." }, "curation": { "has-additional-curation": "yes", "modification-types": [ "other" ], "modification-description": "- Added GEM ID in each sample.\n- Normalize the number of players in each sample with \"N/A\" for consistent data loading.", "has-additional-splits": "no", "additional-splits-description": "N/A", "additional-splits-capacicites": "N/A" }, "starting": { "technical-terms": "- Data-to-text\n- Neural machine translation (NMT)\n- Document-level generation and translation (DGT) ", "research-pointers": "- [Challenges in Data-to-Document Generation](https://aclanthology.org/D17-1239)\n- [Data-to-Text Generation with Content Selection and Planning](https://ojs.aaai.org//index.php/AAAI/article/view/4668)\n- [Findings of the Third Workshop on Neural Generation and Translation](https://aclanthology.org/D19-5601)" } }, "results": { "results": { "other-metrics-definitions": "Model-based measures proposed by (Wiseman et al., 2017):\n- Relation Generation\n- Content Selection\n- Content Ordering", "has-previous-results": "yes", "current-evaluation": "N/A.", "previous-results": "See Table 2 to 7 of (https://aclanthology.org/D19-5601) for previous results for this dataset.", "metrics": [ "BLEU", "ROUGE", "Other: Other Metrics" ], "original-evaluation": "To evaluate the fidelity of the generated content to the input data.", "model-abilities": "- Textual accuracy towards the gold-standard summary.\n- Content faithfulness to the input structured data." } }, "considerations": { "pii": { "risks-description": "- Structured data contain real National Basketball Association player and organization names." }, "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": "Potential overlap of box score tables between splits. This was extensively studied and pointed out by [1].\n\n[1]: Thomson, Craig, Ehud Reiter, and Somayajulu Sripada. \"SportSett: Basketball-A robust and maintainable data-set for Natural Language Generation.\" Proceedings of the Workshop on Intelligent Information Processing and Natural Language Generation. 2020.", "data-unsuited-applications": "Users may interact with a trained model to learn about a NBA game in a textual manner. On generated texts, they may observe factual errors that contradicts the actual data that the model conditions on.\nFactual errors include wrong statistics of a player (e.g., 3PT), non-existent injury information.", "data-discouraged-use": "Publishing the generated text as is. Even if the model achieves high scores on the evaluation metrics, there is a risk of factual errors mentioned above." } }, "context": { "previous": { "is-deployed": "no", "described-risks": "N/A", "changes-from-observation": "N/A" }, "underserved": { "helps-underserved": "no", "underserved-description": "N/A" }, "biases": { "has-biases": "no", "bias-analyses": "N/A", "speaker-distibution": "- English text in this dataset is from Rotowire, originally written by writers at Rotowire.com that are likely US-based.\n- German text is produced by professional translators proficient in both English and German." } } }