Data sets for NLG
This page lists sets of structured data to be used as input for natural language generation tasks, or to inform research on NLG.
Data-to-text/Concept-to-text Generation
These datasets contain data and corresponding texts based on this data.
E2E
http://www.macs.hw.ac.uk/InteractionLab/E2E/#data
Crowdsourced restaurant descriptions with corresponding restaurant data. English.
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.
WeatherGov
https://cs.stanford.edu/~pliang/data/weather-data.zip
Computer-generated weather forecasts from weather.gov (US public forecast), along with corresponding weather data.
WebNLG
https://github.com/ThiagoCF05/webnlg
Crowdsourced descriptions of semantic web entities, with corresponding RDF triples.
Referring Expressions Generation
Referring expression generation is a sub-task of NLG that focuses only on the generation of referring expressions (descriptions) that identify specific entities called targets.
GRE3D3 and GRE3D7: Spatial Relations in Referring Expressions
http://jetteviethen.net/research/spatial.html
Two web-based production experiments were conducted by Jette Viethen under the supervision of Robert Dale. The resulting corpora GRE3D3 and GRE3D7 contain 720 and 4480 referring expressions, respectively. Each referring expression describes a simple object in a simple 3D scene. GRE3D3 scenes contain 3 objects and GRE3D7 scenes contain 7 objects.
RefClef, RefCOCO, RefCOCO+ and RefCOCOg
https://github.com/lichengunc/refer
Referring expressions for objects in images, and the corresponding images.
The REAL dataset
https://datastorre.stir.ac.uk/handle/11667/82
Referring expressions for objects in images, and the corresponding images.
GeoDescriptors
https://gitlab.citius.usc.es/alejandro.ramos/geodescriptors
Geographical descriptions (eg, "Norte de Galicia") and corresponding regions on a map
TUNA Reference Corpus
https://www.abdn.ac.uk/ncs/departments/computing-science/corpus-496.php
https://www.abdn.ac.uk/ncs/documents/corpus.zip [direct download]
The TUNA Reference Corpus is a semantically and pragmatically transparent corpus of identifying references to objects in visual domains. It was constructed via an online experiment and has since been used in a number of evaluation studies on Referring Expressions Generation, as well as in two Shared Tasks: the Attribute Selection for Referring Expressions Generation task (2007), and the Referring Expression Generation task (2008). Main authors: Kees van Deemter, Albert Gatt, Ielka van der Sluis.
COCONUT Corpus
http://www.pitt.edu/~coconut/coconut-corpus.html
http://www.pitt.edu/%7Ecoconut/corpora/corpus.tar.gz [direct download]
COCONUT was a project on “Cooperative, coordinated natural language utterances”. The COCONUT corpus is a collection of computer-mediated dialogues in which two subjects collaborate on a simple task, namely buying furniture. SGML annotations were added according to the COCONUT-DRI coding scheme.
Dialogue Systems
CLASSiC WOZ corpus on InformationPresentation in Spoken Dialogue Systems
http://www.classic-project.org/corpora
CLASSiC is a project on Computational Learning in Adaptive Systems for Spoken Conversation. The Wizard-of-Oz corpus on Information Presentation in Spoken Dialogue Systems contains the wizards' choices on Information Presentation strategy (summary, compare, recommend , or a combination of those) and attribute selection. The domain is restaurant search in Edinburgh. Objective measures (such as dialogue length, number of database hits, number of sentences generated etc.), as well as subjective measures (the user scores) were logged.
Other
PIL: Patient Information Leaflet corpus
http://mcs.open.ac.uk/nlg/old_projects/pills/corpus/PIL/
http://mcs.open.ac.uk/nlg/old_projects/pills/corpus/PIL-corpus-2.0.tar.gz [direct download]
The Patient Information Leaflet (PIL) corpus] is a searchable and browsable collection of patient information leaflets available in various document formats as well as structurally annotated SGML. The PIL corpus was initially developed as part of the ICONOCLAST project at ITRI, Brighton.
Validity of BLEU Evaluation Metric
https://abdn.pure.elsevier.com/en/datasets/data-for-structured-review-of-the-validity-of-bleu
https://abdn.pure.elsevier.com/files/125166547/bleu_survey_data.zip [direct download]
Correlations between BLEU and human evaluations (for MT as well as NLG), extracted from papers in the ACL Anthology
This page was imported semi-automatically from the NLG Resources Wiki which was run by ACL SIGGEN in the years 2005–2009. Please correct conversion errors and help update its contents. Now this page is associated with the Natural Language Generation Portal. |