About

BabelNet 5.0 covers 500 languages and is obtained from the automatic integration of:

• WordNet, the most popular computational lexicon of English (version 3.0).
• WordNet 2020, a fork of the Princeton Wordnet developed under an open source methodology (April 2020 release).
• Wikipedia, the largest collaborative multilingual Web encyclopedia (November 2020 dump).
• OmegaWiki, a large collaborative multilingual dictionary (January 2017 dump).
• Wiktionary, a collaborative project to produce a free-content multilingual dictionary (June 2020 dump).
• Wikidata, a free knowledge base that can be read and edited by humans and machines alike (November 2020 dump).
• GeoNames, a free geographical database covering all countries and containing over eight million placenames (October 2020 dump).
• ImageNet, an image database organized according to the WordNet hierarchy (2011 release).
• Open Multilingual WordNet, a collection of wordnets available in different languages (January 2021): Albanet, Arabic WordNet (AWN v2), BulTreeBank WordNet (BTB-WN), Chinese Open WordNet, Chinese WordNet (Taiwan), Croatian WordNet, DanNet, Greek WordNet, FinnWordNet, Hebrew WordNet, IceWordNet, ItalWordNet, Japanese WordNet, Lithuanian WordNet, Multilingual Central Repository, MultiWordNet, Norwegian WordNet, Open Dutch WordNet, OpenWN-PT, Princeton WordNet, Persian WordNet, plWordNet, Romanian WordNet, Slovak WordNet, sloWNet, Swedish (SALDO), Thai WordNet, WOLF (WordNet Libre du Français), WoNeF, WordNet Bahasa.
• BabelPic, a large collection of non-concrete pictures.
• VerbAtlas, the largest language-independent verb predicate and role resource.
• Translations obtained from sense-annotated sentences.

BabelNet is linked to different resources and applications from the Sapienza NLP group:

• VerbAtlas: a large multilingual verb predicate and role repository.
• InVeRo: intelligible verbs and roles produced by a state-of-the-art neural Semantic Role Labeling system.
• Train-O-Matic: the first large scale silver data creation approach to multilingual Word Sense Disambiguation.
• MuLaN: silver data creation for Word Sense Disambiguation by means of multilingual label propagation.
• OneSeC, SensEmBERT and ARES: latent Transformer-based sense representations which achieve state-of-the-art performance in multilingual Word Sense Disambiguation.
• Conception: human-intelligibile multilingual representations of BabelNet synsets.
• SyntagNet: a large collection of disambiguated free word associations and collocations.
• SyntagRank: a SyntagNet- and BabelNet-based multilingual word sense disambiguation system.
• Babelfy: a multilingual disambiguation and entity linking system.
• Wikipedia Bitaxonomy: a state-of-the-art taxonomy of Wikipedia pages aligned to a taxonomy of Wikipedia categories.

BabelNet is provided as a stand-alone resource with its Java API, a SPARQL endpoint and a Linked Data interface as part of the Linguistic Linked Open Data Cloud (LLOD Cloud) cloud.