[ANN] NERD - Named Entity Recognition and Disambiguation v0.5 released

Dear all,

We would like to announce a major upgrade of NERD - a Named Entity 
Recognition and Disambiguation framework (http://nerd.eurecom.fr). The 
NERD framework unifies the annotation of 10 entity extractors 
(http://nerd.eurecom.fr/documentation#extractors) using the NERD 
ontology (http://nerd.eurecom.fr/ontology) which provides a rich set of 
axioms aligning the taxonomies of these extractors. The NERD ontology is 
composed of two layers: a core layer of 10 main classes, based on a core 
schema used by the NLP community, and an extended layer composed of 75 
classes which correspond to more fine-grained entity types that these 
extractors can recognize.

In this release, we have better aligned the classification provided by 
the extractors. Furthermore, we provide a so-called "combined strategy" 
that enables to launch all extractors while having a conflict resolution 
mechanism. NERD can harmonizes the annotation of documents written in 
English, French, Italian, Spanish, Swedish, Portuguese, Russian, German, 
and Dutch. NERD automatically detects the language of the documents, 
selects and finally harmonizes the annotations of the extractors. The 
NERD framework enables the annotations of web resources, plain texts, 
and SRT timed-text documents so that video subtitles can also be 
annotated. The new User Interfaces features a personalized dashboard 
where a registered user can easily monitor his activity log, and get 
preliminary statistics about its NERD usage.

HOW TO USE NERD?
* The NERD User Interface is at http://nerd.eurecom.fr
* NERD comes also with a REST interface http://nerd.eurecom.fr/api that 
requires an API KEY. You can both use CLI (command line interface) or 
java & python libraries. More details at 
http://nerd.eurecom.fr/documentation.

HOW TO GET YOUR API KEY?
* From the NERD UI, go to the profile item menu (top right corner arrow, 
click on profile). Then fill up the form and click on the <Get Your API 
KEY> button.

ACKNOWLEDGEMENTS:
We would like to especially thank AlchemyAPI, DBpedia Spotlight, 
Extractiv, Lupedia, OpenCalais, Saplo, SemiTags, Wikimeta, Yahoo! 
Content Analysis, and Zemanta and personally people from these 
organizations: Pablo Mendes (from DBpedia Spotlight), Eric Charton (from 
Wikimeta), Shaun Roach (from AlchemyAPI), Andraz Tori (from Zemanta), 
Mattias Tyrberg and Fredrik Horte (from Saplo), Pavel Mihaylov and Marin 
Dimitrov (Ontotext, Lupedia), Nicolas Torzec and Reiner Kraft (from 
Yahoo) with whom we had discussions and received very valuable feedbacks 
on earlier versions of NERD.

Many thanks to the early adopter community of NERD users for your 
feedback and energetic support. We would like to especially thank: 
Sebastian Hellmann, Yunjia Li, Marieke van Erp, Pierre Antoine Champin, 
Olivier Aubert, Ruben Verborgh, Vuk Milicic, and Jose Luis Redondo for 
their contributions to the codebase.

This release of NERD was partly supported by the French National Agency 
through the project OpenSEM - Open Innovation Platform for Semantic 
Media (http://www.opensem.eu/) - and the European Commission through the 
project LinkedTV - Television linked to the Web (http://www.linkedtv.eu/).

Happy harmonizing your annotations!

Cheers,
Giuseppe Rizzo & Raphael Troncy

Received on Friday, 30 November 2012 15:49:53 UTC