REDEN Online
This is the BETA version of a Web tool for named entity disambiguation and linking to the Linked Data cloud.
It takes in input XML-TEI document tagged with named entities, it then ouputs an enriched version of the input file, where these entities are referenced with a link.
The tool also provides two default types of visualization of ouput TEI content thanks to data such as geographic coordinates or pictures automatically retrieved from LD sets.
PLEASE NOTICE that this REDEN ONLINE web interface has only been tested on Google Chrome.
Distribution
The source code of Reden Online can be found on the following repository. The standalone version, i.e. REDEN, is also available on this other repository, and can be configured for handling further use cases.
If you like to test our tool, please find hereafter two taggued TEI files, Guillaume Apollinaire's L’Hérésiarque et Cie (Labex OBVIL) and Albert Thibaudet's Réflexions sur la littérature (Labex OBVIL), the first one can be used to visualize place mentions on a map and the second one to display a gallery of authors' pictures mentioned in the text.
Considerations
- REDEN does not perform named-entity recognition (i.e. detection and classification), instead links already detected named-entity mentions to Linked Data URIs, therefore the TEI input file must already contain the proper TEI tags for annotating named entities using the TEI-XML standard, for instance, placeName for places and persName for persons
- The file needs to be well-formed; REDEN has been tested on TEI files following different schemas, for instance, Obvil's TEIBook schema. Sample TEI file defined according to the aforementioned schema can be found here
- The disambiguation method can decompose the TEI into several portions for the analysis, there are three options, namely, the paragraph, the chapter and the whole text. Further options are possible by modifying the REDEN Online configuration file
- By default, author mentions are linked to URIs of the BnF dataset and place mentions are linked to URIs of the French DBpedia dataset as these data are already configured in the application. It is possible to configure new Linked Data sets using the standalone version
Let's work now!
Copy-paste the content of the TEI file in the input area below
Visualization
The following map shows distribution of places mentioned in the input TEI-XML file, geo-coordinates are obtained via French DBpedia (if chosen in the form).
Visualization
The following gallery shows the important of authors measured as frequency of mention in the input TEI-XML file, data is obtained from Bibliothèque Nationale de France.