Switch branches/tags
Find file History
Permalink
Type Name Latest commit message Commit time
..
Failed to load latest commit information.
links/055-confidence dbpedia links for chapters Apr 27, 2018
README.md updated description Apr 25, 2018

README.md

Links produced with DBpedia Spotlight

This folder contains the produced named entity links for the 2017 article abstracts and for 2017 book chapter abstracts.

This repository is used to produce the data exploiting DBpedia Spotlight framework to discover the named entities. DBpedia Spotlight is a tool for automatically annotating text documents with DBpedia URIs.

The folder contains links where confidence values are 0,55 and support value is 20.

  • Number of distinct links overall: 89.138
  • Total no of distinct links per abstract: 1.646.342
  • No of total found links: 2.841.682

Quick Start

Easiest way to get started exploring these data is to expand the compressed TTL files in the dbpedia-backlinks folder.

Each of these files contain simple mappings between SciGraph documents (articles, chapters) and DBPedia entities. The mappings are encoded in RDF using the schema:mentions predicate.

For example:

<http://scigraph.springernature.com/things/articles/000025e3062fde501f24ad1d106a3b6e#offset_585_597> <http://schema.org/mentions> <http://dbpedia.org/resource/Limit_point> <http://scigraph.springernature.com/things/articles/000025e3062fde501f24ad1d106a3b6e#nlptool=spotlight&confidence=1.0> .
<http://scigraph.springernature.com/things/articles/000025e3062fde501f24ad1d106a3b6e#offset_45_60> <http://schema.org/mentions> <http://dbpedia.org/resource/Inverse_problem> <http://scigraph.springernature.com/things/articles/000025e3062fde501f24ad1d106a3b6e#nlptool=spotlight&confidence=1.0> .
<http://scigraph.springernature.com/things/articles/000025e3062fde501f24ad1d106a3b6e#offset_77_94> <http://schema.org/mentions> <http://dbpedia.org/resource/Signal_processing> <http://scigraph.springernature.com/things/articles/000025e3062fde501f24ad1d106a3b6e#nlptool=spotlight&confidence=0.9999991378390711> .

Credits

This work has been carried out by Beyza Yaman in collaboration with the DBPedia and SciGraph teams, thanks to a research grant from Springer Nature.

A technical article providing more info about this research will be made available soon.