Babelfy v1.0: Multilingual Word Sense Disambiguation and Entity Linking Together!

====================================================================

Babelfy v1.0: Multilingual Word Sense Disambiguation and Entity Linking
together!

http://babelfy.org

====================================================================

As an output of the "MultiJEDI" Starting Grant <http://multijedi.org/>,
funded by the European Research Council

and headed by Prof. Roberto Navigli
<http://wwwusers.di.uniroma1.it/~navigli>, the Linguistic Computing
Laboratory <http://lcl.uniroma1.it/> of the Sapienza

University of Rome is proud to announce the release of Babelfy
<http://babelfy.org/> v1.0.

Babelfy [1,2] is a joint, unified approach to Word Sense Disambiguation and
Entity Linking

in language of choice. The system is based on a loose identification of
candidate meanings

coupled with a densest subgraph heuristic which selects high-coherence
semantic

interpretations. Its performance on standard word sense disambiguation and
entity linking

tasks is on a par with, or surpasses, those of language- and task-specific
state-of-the-art

systems.

Babelfy draws primarily on BabelNet (http://babelnet.org), a very large
encyclopedic

dictionary and semantic network. BabelNet 3.0 covers 271 languages and
provides both

lexicographic and encyclopedic knowledge for all the open-class parts of
speech, including

nearly 14 million concepts and named entities, thanks to the seamless
integration of WordNet,

Wikipedia, Wiktionary, OmegaWiki, Wikidata and the Open Multilingual
WordNet.

New features in Babelfy v1.0:

* 271 languages covered plus a novel language-agnostic setting!

* Available via easy-to-use Java and HTTP RESTful APIs.

* The input context can be either a text or a bag of words where you can mix
up languages!

* Plenty of tunable parameters for the disambiguation procedure such as
setting your own

threshold, enabling multiple scored annotations of the same fragment,
restricting the

annotations to WordNet, Wikipedia or BabelNet, input the offsets that you
want to be linked,

provide pre-annotated tokens as disambiguation context, disable/enable the
most common

sense heuristic, multi-word expressions and the densest subgraph heuristic.

* Three different scores are now output: the disambiguation score, a
coherence score and

a global relevance score.

* Disambiguation and entity linking is performed using BabelNet, thereby
implicitly annotating

according to several different inventories such as WordNet, Wikipedia,
Wiktionary, OmegaWiki,

etc.

[1] Andrea Moro, Alessandro Raganato, Roberto Navigli. Entity Linking meets
Word Sense <http://www.transacl.org/wp-content/uploads/2014/05/54.pdf>

Disambiguation: a Unified Approach
<http://www.transacl.org/wp-content/uploads/2014/05/54.pdf>. Transactions
of the Association for Computational

Linguistics (TACL), 2, pp. 231-244 (2014).

[2] Andrea Moro, Francesco Cecconi, Roberto Navigli. Multilingual Word
Sense Disambiguation

and Entity Linking for Everybody. Proc. of the 13th International Semantic
Web Conference,

Posters and Demonstrations (ISWC 2014), pp. 25-28, Riva del Garda, Italy,
19-23 October 2014.