BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD...
Transcript of BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD...
![Page 1: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/1.jpg)
Roberto Navigli
BabelNet, the LLOD cloud and the Industry
The Luxembourg BabelNet Workshop – Session 4
http://lcl.uniroma1.it
![Page 2: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/2.jpg)
Session 4 – The Luxembourg BabelNet Workshop
[15.45-17.00, 2 March, 2016]
• BabelNet in the Linguistic Linked Open Data (LLOD) cloud
• Industrial applications
• Lightning talks II
03/03/2016BabelNet & Friends - Luxembourg BabelNet Workshop
Roberto Navigli
2
![Page 3: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/3.jpg)
BabelNet, the LLOD cloud and the Industry
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
3
![Page 4: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/4.jpg)
03/03/2016BabelNet & friends
Roberto Navigli
4
![Page 5: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/5.jpg)
The Linked Data cloud
03/03/2016 5BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 6: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/6.jpg)
The Linguistic
Linked Open Data cloud!
![Page 7: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/7.jpg)
RDF-Lemon encoding of BabelNet
• RDF representation based on:
• Dublin Core:
– http://purl.org/dc/terms/ and http://purl.org/dc/elements/1.1/
• lemonhttp://www.lemon-model.net/lemon#
• SKOS (Simple Knowledge Organization System)http://www.w3.org/2004/02/skos/core#
• LexInfo 2.0http://www.lexinfo.net/ontology/2.0/lexinfo#
• BabelNet model:
domain name: http://babelnet.org/rdf/
http://babelnet.org/model/babelnet#
03/03/2016 7BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 8: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/8.jpg)
What is RDF?
• The Resource Description Framework (RDF) is the
W3C standard for encoding knowledge
• A general framework for describing any kind of resource
ranging from concepts to real world objects such as
Web sites, physical devices, etc.
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
8
![Page 9: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/9.jpg)
What is Dublin Core?
• A standard aimed at creating a digital "library card
catalog" for the Web
• Dublin Core provides 15 metadata elements (data that
describes data), including:
– title (the name given to the resource)
– creator (the person or organization who created it)
– subject (the topic covered)
– description (a textual outline of the content)
– publisher (who made the resource available)
– contributor (people who contributed content)
– date (when the resource was made available)
– type (a category for the content)
– language (in what language the content is written)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
9
![Page 10: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/10.jpg)
What is Lemon?
• A model for encoding and structuring lexicons and
machine-readable dictionaries and link them to the
Semantic Web and the (Linguistic) Linked Open Data cloud
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
10
![Page 11: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/11.jpg)
What is LexInfo
• LexInfo is a model for providing linguistic categories for
the Lemon model
• Including:
– Morphological information (tense, person, gender, etc.)
– Part-of-speech tag information
– Syntactic information (subject, directObject, etc.)
– Lexical-semantic relations (antonymy, synonymy, collocations)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
11
![Page 12: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/12.jpg)
What is SKOS?
• Simple Knowledge Organization System (SKOS) is a
specification for thesauri, taxonomies, subject heading
lists, etc.
• Part of the Semantic Web family of standards built upon
RDF
• Objective: easy publication and use of linked data
vocabularies
• Provides concepts (skos:Concept) and basic relations
(skos:related, skos:broader, skos:narrower)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
12
![Page 13: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/13.jpg)
BabelNet in Lemon-RDF
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
13
• the RDF resource consists of a
set of Lexicons, one per
language
• Lexicons gather Lexical Entries
which comprise the forms of an
entry; in our case: words of the
Babel lexicon.
• Lexical Forms encode the
surface realisation(s) of Lexical
Entries; in our case: lemmas of
Babel words.
• Lexical Senses represent the
usage of a word as reference to
a specific concept; in our case:
Babel senses.
• Skos Concepts represent ‘units
of thought’; in our case: Babel
synsets. SKOS concepts
Babel
words
lemmas
Babel
senses
Babel
synsets
![Page 14: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/14.jpg)
@prefix bn: <http://babelnet.org/rdf/>
@preflx lemon: <http://www.lemon-model.net/lemon#>
…
bn:babelNet_lexicon_EN
a Iemon:Lexicon ;
dc:source <http://babelnet.org/>;
lemon:entry bn:Citrus_limon_EN , bn:lemon_EN , bn:Lemon_EN , bn:Lemons_EN , bn:lemon_tree_EN …
lemon:language “EN“.
bn:lemon_tree_EN
a lemon:LexicalEntry ;
rdfs:label “lemon_tree”@EN ;
dc:source <http://wordnet.princeton.edu/> ;
Iemon:canonicalForm <http://babelnet.org/rdf/lemon_tree_n_EN/canonicalForm> ;
lemon:language “EN” ;
lemon:sense <http://babelnet.org/rdf/lemon_tree_EN/s00019309n> ;
lexinfo:partOfSpeech
lexinfo:noun .
<http://babelnet.org/rdf/lemon_tree_n_EN/canonicalForm>
a lemon:Form ;
lemon:writtenRep “lemon_tree”@EN .
<http://babelnet.org/rdf/lemon_tree_EN/s00019309n>
a lemon:sense ;
lemon:reference bn:s00019309n ;
lexinfo:translation <http://babelnet.org/rdf/柠檬树_ZH#00019309n> , <http://babelnet.org/rdf/limoeiro_PT/s00019309n> ,
<http://babelnet.org/rdf/citronnier_FR/s00019309n> , <http://babelnet.org/rdf/lemon_puno_TL/s00019309n> ...
bn:s00019309n
a skos:Concept ;
bn-lemon:synsetType "concept' ;
bn:wikipediaCategory
wikipedia-fr:Catégorie:Fruit_alimentaire , wikipedia-nl:Categorie:Wijnruitfamilie , wikipedia-en: Category:Tropical_agriculture , …;
lemon:isReferenceOf <http://babelnet.org/rdf/lemon_EN/s00019309n> , <http://babelnet.org/rdf/citron_FR/s00019309n> ,
<http://babelnet.org/rdf/Citronensaft_DE/s00019309n> , <http://babelnet.org/rdf/lemon_tree_EN/s00019309n> ...
Iexinfo:membeHolonym
bn:s00037916n ;
skos:exactMatch dbpedia:Lemon , lemon-Omega:OW_eng_Synset_33386 , freebase:m.09k_b , lemon-WordNet31: 112732356-n ;
bn-lemon:definition “Die Zitrone oder Limone ist die etwa faustgroße Frucht des Zitronenbaums aus der Gattung der Zitruspflanzen.”@DE , “A small evergreen tree that originated in Asia
but is widely cultivated for its fruit”@EN ,
“Le citron est un agrume, fruit du citronnier qui a un PH acide de 2,5.”@FR ;
skos:narrower bn: s00019308n ;
skos:related bn:s00188641n , bn:s00052952n , bn:s00047566n …;
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
14
Example
![Page 15: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/15.jpg)
BabelNet in RDF-Lemon
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
15
![Page 16: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/16.jpg)
BabelNet and its links
• 2 billion triples
Links to:
• WordNet-RDF
• Wiktionary
• Apertium
• lemonUby
• Lexinfo
• DBpedia
• Zhishi.lemon
• YAGO
• Freebase
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
16
![Page 17: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/17.jpg)
Online availability of BabelNet RDF
• SPARQL endpoint
– virtuoso universal server
– http://babelnet.org/sparql/
• Dereferencing
– http://babelnet.org/rdf/
– Pubby Linked Data Frontend
(http://wifo5-03.informatik.uni-mannheim.de/pubby/)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
17
![Page 18: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/18.jpg)
Let's see some examples of SPARQL queries
• Open a browser
• Enter the following URI: http://babelnet.org/sparql/
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
18
![Page 19: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/19.jpg)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
19
Example 1: Retrieve the senses of a given lemma
• Given a word, e.g. home, retrieve all its senses and
corresponding synsets in all supported languages:
SELECT DISTINCT ?sense, ?synset WHERE {
?entries a lemon:LexicalEntry .
?entries lemon:sense ?sense .
?sense lemon:reference ?synset .
?entries rdfs:label "home"@en .
}
![Page 20: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/20.jpg)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
20
Example 1: Retrieve the senses of a given lemma
• Given a word, e.g. home, retrieve all its senses and
corresponding synsets in all supported languages:
SELECT DISTINCT ?sense, ?synset WHERE {
?entries a lemon:LexicalEntry .
?entries lemon:sense ?sense .
?sense lemon:reference ?synset .
?entries rdfs:label "casa"@it .
}
![Page 21: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/21.jpg)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
21
Example 2: Retrieve the translation of a given sense
• Given the URI of a given word sense:
SELECT ?translation WHERE {
?entry a lemon:LexicalSense .
?entry lexinfo:translation ?translation .
FILTER
(?entry=<http://babelnet.org/rdf/house_EN/s00044994n>)
}
![Page 22: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/22.jpg)
Example 3: retrieve the license information about a
given sense
SELECT ?license WHERE {
?entry a lemon:LexicalSense .
?entry dcterms:license ?license .
FILTER
(?entry=<http://babelnet.org/rdf/Bill_Gates_%28Mi
crosoft%29_EN/s00010401n>)
}
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
22
![Page 23: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/23.jpg)
Example 4: retrieve textual definitions for a synset
SELECT DISTINCT ?language ?gloss ?license ?sourceurl
WHERE {
?url a skos:Concept .
?url bn-lemon:synsetID ?synsetID .
OPTIONAL {
?url bn-lemon:definition ?definition .
?definition lemon:language ?language .
?definition bn-lemon:gloss ?gloss .
?definition dcterms:license ?license .
?definition dc:source ?sourceurl .
}
FILTER (?url=<http://babelnet.org/rdf/s03083790n>)
}03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
23
![Page 24: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/24.jpg)
Industrial applications
• Concept and Named Entity Extraction
• Dictionary of the future
• New publishing initiatives
• Computer-Assisted Translation
• Multilingual Text Analytics
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
24
![Page 25: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/25.jpg)
Industrial application:
Concept and Named Entity Extraction
Who is interested: any company interested in extracting
knowledge from their document base
![Page 26: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/26.jpg)
Concept and Named Entity Extraction
• Compared to terminology extraction, which extracts
only terms, here we extract concepts and named
entities
• With BabelNet:
– in any language
– making the outputs comparable across languages
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
26
![Page 27: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/27.jpg)
Concept and Named Entity Extraction: an example
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
27
![Page 28: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/28.jpg)
Concept and Named Entity Extraction: an example
• Key terms extracted with statistical techniques:
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
28
![Page 29: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/29.jpg)
• Key conceptual categories:
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
29
Concept and Named Entity Extraction: an example
![Page 30: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/30.jpg)
Concept and Named Entity Extraction: an example
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
30
![Page 31: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/31.jpg)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
31
Concept and Named Entity Extraction: an example
![Page 32: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/32.jpg)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
32
Concept and Named Entity Extraction: an example
![Page 33: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/33.jpg)
Industrial application:
The Dictionary of the Future
Who is interested: all producers of language
resources
BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 34: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/34.jpg)
Today's (Paper) Dictionaries
• They are not hypertexts
• Organized alphabetically (except for
thesauri and analogical dictionaries)
• Not "browsable" by semantic
correlation, synonymy or other criteria
03/03/2016 34BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 35: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/35.jpg)
Machine-readable dictionaries
• Machine-readable dictionaries as
conversions of paper dictionaries
that connect words via hyperlinks
• Problem 1: these links are not
"disambiguated", as they do not
take you to the appropriate meaning
of the word
– Es. If I click on sveglio (meaning both
awaken and apt, smart), I will be taken to
the entry of that word, but not on its
most suitable meaning.
03/03/2016 35BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 36: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/36.jpg)
Dictionaries are like Islands without Bridges!
• Problem 2: monolingual and bilingual dictionaries are
disconnected from each other
• But we could use the source language as a pivot to
connect translations across languages (e.g. English-
Italian, English-French, English-Chinese, etc.), so as to
produce multilingual entries, like in BabelNet
• This requires advanced disambiguation techniques
03/03/2016Multilingual Web Access – WWW 2015
Roberto Navigli
36
![Page 37: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/37.jpg)
What can Language Resource Producers do?
1. Disambiguate translations of their bilingual
dictionaries, so as to create a browsable semantic
network
2. Similarly with phrases and example sentences
03/03/2016 37
apt2
sveglio1
pronto1
clever2
brainy1
quick3
bright5
intelligente1
perspicace1
acuto1
BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 38: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/38.jpg)
• Connect source language meanings across languages in
the different bilingual dictionaries, obtaining a multilingual
dictionary and network
• Beyond innovative browsing experience for the user, you
can identify missing meanings or potential mistakes03/03/2016 38
apt2
sveglio1
pronto1
clever2
brainy1
quick3
bright5机敏
聪明
智能
BabelNet, the LLOD cloud and the Industry
Roberto Navigli
What can Language Resource Producers do?
![Page 39: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/39.jpg)
• Connect different monolingual dictionaries and thesauri,
thus creating a richer and unified experience for the user
– Think of thousands of glossaries on the Web and EU resources
• The user does not have to choose between a traditional
dictionary, an analogical dictionary and a thesaurus, but
will find all the information aggregated in a single entry
(but keeping the links back to the respective resources)
• Definition: Il più comune degli aeromobili a sostentazione
dinamica
• Synonyms: aeroplano, aereo, apprecchio
• Actions: pilotare, volare con, imbarcare, sbarcare, ecc.
• Collocations: a due piani, invisibile, canadair, ecc.
03/03/2016 39BabelNet, the LLOD cloud and the Industry
Roberto Navigli
What can Language Resource Producers do?
![Page 40: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/40.jpg)
Industrial application:
New Publishing Initiatives
Who is interested: all producers of
educational and technical multimedia content
BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 41: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/41.jpg)
New Publishing Initiatives
• Annotate text with concepts, definitions and images
• School texts, also in multiple languages
03/03/2016 41BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 42: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/42.jpg)
Consider the semantic annotation of free text
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
42
![Page 43: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/43.jpg)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
43
Consider the semantic annotation of free text
![Page 44: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/44.jpg)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
44
Consider the semantic annotation of free text
![Page 45: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/45.jpg)
Example:
You are running a booking or reviewing web site
Assume you want to semantically analyze your reviews:
• "Una serata stupenda all'insegna della cucina romana” -Recensito 2
settimane fa
• E' impressionante come questo locale mantenga elevatissima la
qualità e la freschezza degli ingredienti, grande la passione nella
cucina e prezzi assolutamente competitivi! Dei tantissimi locali da
me visitati fino a oggi, questo ha di gran lunga il miglior rapporto
qualità prezzo a Roma (e, oserei dire, tra i migliori in Italia). E' vero,
ogni tanto non hanno il menù, ma il conto è, ieri come sempre,
impeccabile (tra i 25 e i 30 euro a testa per due primi di paccheri
carciofi e guanciale, un secondo di pesce - un'ombrina freschissima
- e 4 contorni di verdure di stagione, tra cui carciofi, broccoletti,
cicoria). Fantastici! Continuate così!
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
45
![Page 46: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/46.jpg)
What can you do?
• People can search by dish independently of their
language
– What restaurants serve spaghetti with cheese and pepper?
– Quels restaurants servent des artichauts?
• People can search by similarity
– Give me all the restaurants that cook dishes containing
ingredients similar to paccheri al pomodoro?
• People can have explanations of the ingredients or
dishes cooked
• People can browse and explore dishes
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
46
![Page 47: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/47.jpg)
Same can be done with unstructured tags
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
47
![Page 48: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/48.jpg)
Same can be done with unstructured tags
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
48
![Page 49: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/49.jpg)
Same can be done with unstructured tags
• Because Babel synsets are multilingual, I can now
search the database in any language
– For instance, programmazione instead of programming
• I can exploit the semantic network structure, to find
people who are semantically closest to my interests or
to skills I am looking for
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
49
![Page 50: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/50.jpg)
Industrial application:
Computer-Assisted Translation
Who is interested: all producers of computer-
assisted translation software
BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 51: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/51.jpg)
Computer-Assisted Translation (CAT)
03/03/2016Natural Language Processing: An Introduction
Roberto Navigli
Pagina 51
![Page 52: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/52.jpg)
Computer Assisted Translation (CAT)
03/03/2016Natural Language Processing: An Introduction
Roberto Navigli
Pagina 52
![Page 53: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/53.jpg)
03/03/2016BabelNet, the LLOD cloud and the Industry
Roberto Navigli
53
• From XTM's system:
Computer Assisted Translation (CAT)
![Page 54: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/54.jpg)
Industrial application:
Multilingual Text Analytics
Who is interested: media content providers
and analysts, etc.
BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 55: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/55.jpg)
03/03/2016Multilingual Web Access – WWW 2015
Roberto Navigli
55
![Page 56: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/56.jpg)
Trends are ambiguous
• The senses of the word “mercury” are conflated and
plotted together.
03/03/2016 56BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 57: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/57.jpg)
Dream: multilingual semantic analytics
• Goal: semantic analytics of text in any language for
both named entities and concepts
03/03/2016 57BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 58: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/58.jpg)
Multilingual News Analytics and Comparison
03/03/2016 58
• Extract the most important concepts and Named Entities
in virtually any language (271 covered currently)
BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 59: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/59.jpg)
03/03/2016 59
• Important semantic common ground
• But also: complementarity!
BabelNet, the LLOD cloud and the Industry
Roberto Navigli
![Page 60: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/60.jpg)
03/03/2016BabelNet & friends
Roberto Navigli
60
![Page 61: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/61.jpg)
Thanks or…
m i(grazie)
6103/03/2016BabelNet, Babelfy and Beyond!
Roberto Navigli
![Page 62: BabelNet, the LLOD cloud and the · PDF filelexinfo:translation , ... BabelNet, the LLOD cloud and the Industry 03/03/2016 39 Roberto Navigli](https://reader031.fdocuments.us/reader031/viewer/2022030410/5a974be67f8b9a9c5b8d2e7b/html5/thumbnails/62.jpg)
Roberto Navigli
Linguistic Computing Laboratory
http://lcl.uniroma1.it
@RNavigli