Data sets for NLG - Revision history

Wangqy96: /* Data-to-text/Concept-to-text Generation */

2021-05-31T04:03:52Z

Data-to-text/Concept-to-text Generation

← Older revision		Revision as of 04:03, 31 May 2021
Line 85:		Line 85:

	Company descriptions collected from Wikipedia. The dataset contains semantic representations, short, and long descriptions for 51K companies in English		Company descriptions collected from Wikipedia. The dataset contains semantic representations, short, and long descriptions for 51K companies in English

			=== Wikipedia Pre-train Pairs Dataset ===
			https://eaglew.github.io/dataset/stage

			This repository contains 542,192 data pairs used for the Wikipedia fine-tuning stage. The data folder contains 166 JSON files which include graph-to-text pairs related to 15 categories (Astronaut, University, Monument, Building, ComicsCharacter, Food, Airport, SportsTeam, WrittenWork, Athlete, Artist, City, MeanOfTransportation, CelestialBody, Politician) that appear in the WebNLG dataset.

	== Referring Expressions Generation==		== Referring Expressions Generation==

Wangqy96: /* ReviewRobot Dataset */

2020-10-31T00:03:42Z

ReviewRobot Dataset

← Older revision		Revision as of 00:03, 31 October 2020
Line 230:		Line 230:
	https://github.com/EagleW/ReviewRobot/blob/master/dataset/README.md		https://github.com/EagleW/ReviewRobot/blob/master/dataset/README.md

	This dataset contains 8,110 paper and review pairs and background KG from 174,165 papers. It also contains information extraction results from SciIE ~~and~~ various knowledge graphs built on the IE results.		This dataset contains 8,110 paper and review pairs and background KG from 174,165 papers. It also contains information extraction results from SciIE, various knowledge graphs built on the IE results, and human annotation for paper-review pairs.

	==Challenge Data Repository ==		==Challenge Data Repository ==

Wangqy96: /* ReviewRobot Dataset */

2020-10-30T17:42:55Z

ReviewRobot Dataset

← Older revision		Revision as of 17:42, 30 October 2020
Line 228:		Line 228:

	=== ReviewRobot Dataset ===		=== ReviewRobot Dataset ===
	https://github.com/EagleW/ReviewRobot/blob/~~main~~/dataset/README.md		https://github.com/EagleW/ReviewRobot/blob/master/dataset/README.md

	This dataset contains 8,110 paper and review pairs and background KG from 174,165 papers. It also contains information extraction results from SciIE and various knowledge graphs built on the IE results.		This dataset contains 8,110 paper and review pairs and background KG from 174,165 papers. It also contains information extraction results from SciIE and various knowledge graphs built on the IE results.

Wangqy96: /* Paper Generation */

2020-10-30T17:32:37Z

Paper Generation

← Older revision		Revision as of 17:32, 30 October 2020
Line 229:		Line 229:
	=== ReviewRobot Dataset ===		=== ReviewRobot Dataset ===
	https://github.com/EagleW/ReviewRobot/blob/main/dataset/README.md		https://github.com/EagleW/ReviewRobot/blob/main/dataset/README.md

	This dataset contains 8,110 paper and review pairs and background KG from 174,165 papers. It also contains information extraction results from SciIE and various knowledge graphs built on the IE results.		This dataset contains 8,110 paper and review pairs and background KG from 174,165 papers. It also contains information extraction results from SciIE and various knowledge graphs built on the IE results.

Wangqy96: /* Paper Generation */

2020-10-30T17:32:26Z

Paper Generation

← Older revision		Revision as of 17:32, 30 October 2020
Line 226:		Line 226:

	This dataset gathers three types of pairs: Title-to-Abstract (Training: 22,811/Development: 2095/Test: 2095), Abstract-to-Conclusion and Future work (Training: 22,811/Development: 2095/Test: 2095), Conclusion and Future work-to-Title (Training: 15,902/Development: 2095/Test: 2095) from PubMed. Each pair contains a pair of input and output as well as the corresponding terms(from original KB and link prediction results).		This dataset gathers three types of pairs: Title-to-Abstract (Training: 22,811/Development: 2095/Test: 2095), Abstract-to-Conclusion and Future work (Training: 22,811/Development: 2095/Test: 2095), Conclusion and Future work-to-Title (Training: 15,902/Development: 2095/Test: 2095) from PubMed. Each pair contains a pair of input and output as well as the corresponding terms(from original KB and link prediction results).


	=== ReviewRobot Dataset ===		=== ReviewRobot Dataset ===

Wangqy96: /* Paper Generation */

2020-10-30T17:32:11Z

Paper Generation

← Older revision		Revision as of 17:32, 30 October 2020
Line 226:		Line 226:

	This dataset gathers three types of pairs: Title-to-Abstract (Training: 22,811/Development: 2095/Test: 2095), Abstract-to-Conclusion and Future work (Training: 22,811/Development: 2095/Test: 2095), Conclusion and Future work-to-Title (Training: 15,902/Development: 2095/Test: 2095) from PubMed. Each pair contains a pair of input and output as well as the corresponding terms(from original KB and link prediction results).		This dataset gathers three types of pairs: Title-to-Abstract (Training: 22,811/Development: 2095/Test: 2095), Abstract-to-Conclusion and Future work (Training: 22,811/Development: 2095/Test: 2095), Conclusion and Future work-to-Title (Training: 15,902/Development: 2095/Test: 2095) from PubMed. Each pair contains a pair of input and output as well as the corresponding terms(from original KB and link prediction results).


			=== ReviewRobot Dataset ===
			https://github.com/EagleW/ReviewRobot/blob/main/dataset/README.md
			This dataset contains 8,110 paper and review pairs and background KG from 174,165 papers. It also contains information extraction results from SciIE and various knowledge graphs built on the IE results.

	==Challenge Data Repository ==		==Challenge Data Repository ==

Ereiter: add SportSett

2020-10-21T14:12:25Z

add SportSett

← Older revision		Revision as of 14:12, 21 October 2020
Line 12:		Line 12:
	These datasets contain data and corresponding texts based on this data.		These datasets contain data and corresponding texts based on this data.

	=== boxscore-data (Rotowire) ===		=== boxscore-data (Rotowire) and SportSett ===
	https://github.com/harvardnlp/boxscore-data/		https://github.com/harvardnlp/boxscore-data/

	~~This dataset~~ consists of (human-written) NBA basketball game summaries aligned with their corresponding box- and line-scores.		Boxscore-data consists of (human-written) NBA basketball game summaries aligned with their corresponding box- and line-scores.

			https://github.com/nlgcat/sport_sett_basketball

			SportSett is an expanded data set which includes additional information about basketball games. It is structured as a relational DB

	=== E2E ===		=== E2E ===

Raojinfeng at 20:00, 6 August 2020

2020-08-06T20:00:25Z

← Older revision		Revision as of 20:00, 6 August 2020
Line 46:		Line 46:

	100,000 examples of descriptions of the content of highlighted cells in a Wikipedia table.		100,000 examples of descriptions of the content of highlighted cells in a Wikipedia table.

			=== Weather ===
			https://github.com/facebookresearch/TreeNLG

			~30K human annotated utterances for tree-structured weather meaning representations.

	=== WeatherGov ===		=== WeatherGov ===

Wangqy96: /* Paper Generation */

2020-08-02T19:17:11Z

Paper Generation

← Older revision		Revision as of 19:17, 2 August 2020
Line 217:		Line 217:

	This dataset gathers three types of pairs: Title-to-Abstract (Training: 22,811/Development: 2095/Test: 2095), Abstract-to-Conclusion and Future work (Training: 22,811/Development: 2095/Test: 2095), Conclusion and Future work-to-Title (Training: 15,902/Development: 2095/Test: 2095) from PubMed. Each pair contains a pair of input and output as well as the corresponding terms(from original KB and link prediction results).		This dataset gathers three types of pairs: Title-to-Abstract (Training: 22,811/Development: 2095/Test: 2095), Abstract-to-Conclusion and Future work (Training: 22,811/Development: 2095/Test: 2095), Conclusion and Future work-to-Title (Training: 15,902/Development: 2095/Test: 2095) from PubMed. Each pair contains a pair of input and output as well as the corresponding terms(from original KB and link prediction results).




	==Challenge Data Repository ==		==Challenge Data Repository ==

Wangqy96 at 19:16, 2 August 2020

2020-08-02T19:16:51Z

← Older revision		Revision as of 19:16, 2 August 2020
Line 205:		Line 205:

	Improved annotations for the QGSTEC corpus (with higher inter-rater reliability) as described in [http://oro.open.ac.uk/47284/ Godwin and Piwek (2016)].		Improved annotations for the QGSTEC corpus (with higher inter-rater reliability) as described in [http://oro.open.ac.uk/47284/ Godwin and Piwek (2016)].

			== Paper Generation ==

			=== ACL Title and Abstract Dataset ===
			https://github.com/EagleW/ACL_titles_abstracts_dataset

			This dataset gathers 10,874 title and abstract pairs from the ACL Anthology Network (until 2016).

			=== PubMed Term, Abstract, Conclusion, Title Dataset ===
			https://eaglew.github.io/dataset/paperrobot_writing

			This dataset gathers three types of pairs: Title-to-Abstract (Training: 22,811/Development: 2095/Test: 2095), Abstract-to-Conclusion and Future work (Training: 22,811/Development: 2095/Test: 2095), Conclusion and Future work-to-Title (Training: 15,902/Development: 2095/Test: 2095) from PubMed. Each pair contains a pair of input and output as well as the corresponding terms(from original KB and link prediction results).




	==Challenge Data Repository ==		==Challenge Data Repository ==