Resources

Datasets

• WIKIBIO
  • Link: https://github.com/DavidGrangier/wikipedia-biography-dataset
  • This dataset contains 728,321 biographies from Engish Wikipedia.
• WIKIBIO (French and German)
  • Link: https://github.com/PrekshaNema25/StructuredData_To_Descriptions
  • French and German version of WikiBio dataset.
  • It contains 170K and 50K examples and the vocabulary size was 297K and 143K for French and German respectively.
• WEBNLG
  • http://webnlg.loria.fr/pages/challenge.html
  • Dataset consists of 21,855 data/text pairs of which 8,372 distinct data input. The input contains entities belonging to 9 distinct DBpedia categories (Astronaut, University, Monument, Building, ComicsCharacter, Food, Airport, SportsTeam and WrittenWork).
• E2E
  • Link: http://www.macs.hw.ac.uk/InteractionLab/E2E/
  • Dataset for MR to text task which contains 50k instances belonging to restaurant domain.
• AMR-to-English Generation
  • http://alt.qcri.org/semeval2017/task9/
• WIKITABLETEXT
• SBNATION
  • https://github.com/harvardnlp/boxscore-data
  • Dataset contains 10903 distinct rotowire summaries.
• ROTOWIRE
  • https://github.com/harvardnlp/boxscore-data
  • Dataset contains 4853 distinct rotowire summaries.
• WEATHERGOV
  • https://cs.stanford.edu/~pliang/data/weather-data.zip
  • 3,753 instances of weather forecast table and corresponding summaries. Each table contains 36 records along with alignments.
• ROBOCUP
• ATIS dataset
  • http://www.ikonstas.net/index.php?page=resources
  • Dataset is of Airtravel domain contains 4962 training and 448 testing examples.
• WikiTablePara
  • https://github.com/parajain/structscribe
  • multi-domain benchmark dataset containing WikiTables and their descriptions
• SUMTIME
  • https://ehudreiter.files.wordpress.com/2016/12/sumtime.zip
  • Weather forecasts written by human forecasters, with corresponding forecast data, for UK North Sea marine forecasts.

Systems

• Heuristic driven NLG Systems
• GenL (http://kowey.github.io/GenI) :
• RealPro
• GoPhi
• KPML
• RNNLG (https://shawnwun.github.io/talks/DL4NLG_20160906.pdf)
• Other NLG systems: https://aclweb.org/aclwiki/Downloadable_NLG_systems