DBpedia Spotlight

Links

website - http://www.dbpedia-spotlight.org

demo service - http://demo.dbpedia-spotlight.org

General Notes

Since v1.0, DBpedia Spotlight was split into two versions, under the same API, as follow:

DBpedia-Spotlight-Model: described in Improving Efficiency and Accuracy in Multilingual Entity Extraction
DBpedia-Spotlight-Lucene: described in DBpedia Spotlight: Shedding Light on the Web of Documents

This important movement was the way that we found to deliver faster fixes and new releases, providing solutions for each annotation approach.

Our first achievement is related with licensing. DBpedia Spotlight Model is now full compliance with Apache 2.0. It means that you can use it without any commercial restrictions.

We are so excited because there's even more great news to come.

If you require any further information, feel free to contact us via [email protected]. We are already very excited to spend time with you on further community meetings and to publish new DBpedia releases.

Keep annotating,

All the best

Shedding Light on the Web of Documents

DBpedia Spotlight looks for ~3.5M things of unknown or ~320 known types in text and tries to link them to their global unique identifiers in DBpedia.

Compilation

The requirements to compile the DBpedia Spotlight project are as follows:

Maven 3.1, or later.
JDK 1.8

The process to compile the DBpedia Spotlight code to produce the corresponding jar files are as follows:

[optional] export JAVA_HOME=/path/of/the/jdk1.8/ - only required if the default JDK is different
run mvn package - this process will take some time

Each module (core, index, rest) will produce the corresponding jar file in the "target" directory, e.g., core/target or index/target. The jar file containing the DBpedia Spotlight annotation service will be in the rest module directory (rest/target/rest-1.1-jar-with-dependencies.jar).

To install the DBpedia Spotlight in the local maven repository, replace the mvn package command with the mvn install command.

Testing the DBpedia Spotlight service (the compiled jar file)

To test the produced jar file, download a language model from the DBpedia Databus, uncompress it and execute the following command:

java -Dfile.encoding=UTF-8 -Xmx10G -jar rest-1.1-jar-with-dependencies.jar /path/to/the/uncompress/language/model http://0.0.0.0:2222/rest

When the DBpedia Spotlight annotation service is ready, it could be queried through a curl command as follows:

curl http://0.0.0.0:222/rest/annotate --data-urlencode "text=Barack Obama was the president of the United States" - H "Accept: text/turtle"

Demonstration

Go to our Demonstration page, copy+paste some text and play with the parameters to see how it works.

Endpoints

https://api.dbpedia-spotlight.org/{LANGUAGE}/annotate

English: https://api.dbpedia-spotlight.org/en/annotate
German: https://api.dbpedia-spotlight.org/de/annotate
Dutch: https://api.dbpedia-spotlight.org/nl/annotate
French: https://api.dbpedia-spotlight.org/fr/annotate
Italian: https://api.dbpedia-spotlight.org/it/annotate
Russian: https://api.dbpedia-spotlight.org/ru/annotate
Spanish: https://api.dbpedia-spotlight.org/es/annotate
Portuguese: https://api.dbpedia-spotlight.org/pt/annotate
Hungarian: https://api.dbpedia-spotlight.org/hu/annotate
Turkish: https://api.dbpedia-spotlight.org/tr/annotate

Call our web service

You can use our demonstration Web Service directly from your application.

curl https://api.dbpedia-spotlight.org/en/annotate  \
  --data-urlencode "text=President Obama called Wednesday on Congress to extend a tax break
  for students included in last year's economic stimulus package, arguing
  that the policy provides more generous assistance." \
  --data "confidence=0.35"

or for JSON:

curl https://api.dbpedia-spotlight.org/en/annotate  \
  --data-urlencode "text=President Obama called Wednesday on Congress to extend a tax break
  for students included in last year's economic stimulus package, arguing
  that the policy provides more generous assistance." \
  --data "confidence=0.35" \
  -H "Accept: application/json"

Run your own server

If you need service reliability and lower response times, you can run DBpedia Spotlight in your own In-House Server based on Docker.

Models and data

Models and raw data for most languages are available here.

Citation

If you use DBpedia Spotlight in your research, please cite the following paper:

@inproceedings{isem2013daiber,
  title = {Improving Efficiency and Accuracy in Multilingual Entity Extraction},
  author = {Joachim Daiber and Max Jakob and Chris Hokamp and Pablo N. Mendes},
  year = {2013},
  booktitle = {Proceedings of the 9th International Conference on Semantic Systems (I-Semantics)}
}

Licenses

All the original code produced for DBpedia Spotlight Model is licensed under Apache License, 2.0.

dbpedia-spotlight/dbpedia-spotlight-model