Welcome to the Linked Data server for NELL

This server publishes Linked Data from the NELL ontology and knowledge base.
The last version is 0.3#1100.
Versions 0.2#xxxx upwards are not only kept up-to-date with latest NELL versions, but they also provide rich provenance metadata using different provenance representations.

Extended information can be found in the associated technical report.

Downloads 0.3#1100

Datasets provided under the CC0 1.0 license.
NELL ontology can be downloaded here.
NELL2RDF metadata ontology can be downloaded here.
You can download NELL2RDF data in different flavours, whether as HDT dumps or gziped N-Triples dumps.

[ w/o provenance metadata ]
Turtle (315 MB)* GZ (Zipped 309 MB)
[ w/ provenance modeled using RDF reification ]
Turtle (13.4 GB)* HDT (Zipped 10.2 GB)
[ w/ provenance modeled using n-ary relations ]
Turtle (13.4 GB)* HDT (Zipped 10.1 GB)
[ w/ provenance modeled using Named Graphs ]
N/A HDT (Zipped 9.9 GB)
[ w/ provenance modeled using the Singleton Property ]
Turtle (19 GB)* HDT (Zipped 10.4 GB)
[ w/ provenance modeled using NDFluents ]
Turtle (14.0 GB)* HDT (Zipped 10.4 GB)

*HDT is a highly optimised compression format for RDF.

Interested users are encouraged to extend NELL2RDF further. We are always available for any assistance.


The Never-Ending Language Learning (aka NELL) is a project that is part of the Read-The-Web initiative conducted at Carnegie Mellon University. NELL is running since January 2010. It visits iteratively Web pages, and extract knowledge out of the unstructured data present in the source page.

SPARQL endpoints

You can access a SPARQL enpoint for any of these datasets here (you will need to log in with user "anonymous", password "anonymous").

Source code

Source code for version 0.3 software is open source on github.
Software is provided under the LGPL v3.0.

Older Dumps

Previous data dumps and ontologies are available in our repository.


When citing v0.3#xxxx, please use the following :

José M. Giménez-García, Maísa Duarte, Antoine Zimmermann, Christophe Gravier, Estevam R. Hruschka Jr., Pierre Maret, "Nell2RDF: eading the Web, and Publishing it as Linked Data"

When citing v0.1#690, please use the following :

A. Zimmermann, C. Gravier, J. Subercaze, and Q. Cruzille, "Nell2RDF: Read the Web, and turn it into RDF", 2nd Int Workshop on Knowledge Discovery and Data Mining Meets Linked Open Data, 10th ESWC 2013, Montpellier, France


The version of the software tells you two things: (1) the version of the application used to generate RDF from NELL's dataset, and (2) the last NELL iteration used for the generation. For example, in the present case, 0.3#1100 tells you that we are running version 0.3 of our software with NELL iteration number 1100.

List of contributors :

José M. Giménez-García, Antoine Zimmermmann, Christophe Gravier, Maísa Duarte, Estevam R. Hruschka Jr., Pierre Maret, Julien Subercaze, Quentin Cruzille.

This work is partly supported by the WDAQua research and innovation program, Marie Sklodowska-Curie grant agreement No 642795.

This work has also been supported by Université Jean Monnet and Ecole Supérieure des Mines de Saint-Etienne.