Upload
semanticsconference
View
48
Download
0
Embed Size (px)
Citation preview
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 1
Co-funded by the Horizon 2020Framework Programme of the European UnionGrant Agreement Number 644771
FREME: Putting Standards into action
FREME Tutorial | Leipzig, 12 September 2016
Phil Ritchie (1), with contributions from Felix Sasaki (2)(1) Vistatec (2) DFKI / W3C Fellow
www.freme-project.eu
FREME to make linked data available to Localizers – FREME at FEISGILTT 2015 WWW.FREME-PROJECT.EU 2
CURRENT STATE OF SOLUTIONS
Machine translation, terminology
annotation, ...
Linked data creation & processing
GAPS THAT HINDER BUSINESS:
• Plethora of formats• Adaptability and platform dependency• Language coverage• Usability “The right tool for the right person
in given and new enterprises”: technology influences job profiles
FREME to make linked data available to Localizers – FREME at FEISGILTT 2015 WWW.FREME-PROJECT.EU 3
FREME TO THE RESCUE: ENRICHING DIGITAL CONTENT
Machine translation, terminology
annotation, ...
Linked data creation & processing
LT and LD as first class citizens on the Web
A SET OF INTERFACES* - DESIGN DRIVENBY BUSINESS CASES
LT and LD for varioususer types: (application) developer, content architect, content author, …
* Graphical interfaces* Software Interfaces
THE FREME TUTORIAL WWW.FREME-PROJECT.EU 4
THE FREME PROJECT
• Two year H2020 Innovation action; start February 2015
• Industry partners leading four business cases arounddigital content and (linked) data
• Technology development bridging language and data
• Outreach and business modelling demonstrating monetization of the multilingual data value chain
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 5
The FREME Framework
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 6
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 7
Before FREME
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 8
FREME Architecture
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 9
Standards in FREME – why? High level pitch:
• Ease re-use of components & adaptability• E.g. with regards to content round-tripping
• No need to hard-wire linked data and language technology workflows• Named entity recognition followed by
machine translation – or the other way round• Standards and open source
• Provide reference implementations• Have a R&D and business community growing
decentralised & organically – not top down
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 10
Standards and open source: Some details for FREME• NIF (Natural Language Processing Interchange
Format)• Pivot format for enrichment workflows
• ITS (Internationalization Tag Set) 2.0• Metadata for NLP workflows
• OntoLex lemon• Lexica represented as linked data
• Okapi framework• Content round tripping for many formats
• FREME framework itself!• Available under Apache 2.0 license
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 11
What is Internationalization
• Internationalization is a best practice.• It prepares an item that is destined for consumption in a
geographic locale different to its origin for the processes which will adapt it for those target geographic regions
• For content this typically means◦ Separating file structure from content◦ Identifying translatable and non-translatable text
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 12
What does e-Internationalization do?
• e-Internationalization is a utility service• It filters the content format to identify translatable content and
then converts it into FREME’s lingua-franca which is NIF.• Thus it makes FREME enrichment accessible to many users
who would not have the ability or time to make such a conversion
• Built on top of Okapi Framework• Uses sub-set of features of Internationalization Tag Set 2.0 with
categories under itsrdf: namespace.
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 13
Incremental enrichment with NIF
Defines classes of Document, Paragraph, Phrase, Sentence and Word
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 14
Internationalization Tag Set 2.0
ITS 2.0
TextAnalytics
/ Dis-
ambiguation
Translate /
Localisation Note
Domain /MT
Confidence
Quality
Provenance
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 15
ITS RDF Ontology
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 16
XML Localization Interchange File Format
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 17
Community standards versus formal standards?• No difference with regards to usefulness of
standard• No formal standard: can be a big challenge
with regards to adoption by big companies• Royalty free standards are key for long term
adoption!
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 18
skeleton
The Round Trip Process
skeleton content content
FREME enrichmentservice
content
FREMEe-Internationalization
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 19
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 20
RDF in XLIFF
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 21
W3C RDF and XML INTEROPERABILITY COMMUNITY
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 22
Two Business Cases
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 23Wripl Content Strategyzer
NIF AND ITS ROLE IN FREME WWW.FREME-PROJECT.EU 24
Demonstration
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 25Open and document and enrichment starts in background: term & entity spotting; translation…
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 26As enrichment completes, machine translations are available
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 27You can view the enrichments of the source language
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 28All identified terminological concepts and named entities are shown and can be selected…
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 29When clicked, the entity resources can be seen and when clicked again…
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 30SPARQL queries are executed to retrieve desired related information
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 31If a machine translation suggestion is chosen the enrichment pipeline is executed on the translation
FREME: Putting Standards into Action – SEMANTiCS 2016 WWW.FREME-PROJECT.EU 32
Phil Ritchie
CTO, Vistatec
Twitter: @philinthecloud
Photo
[to be included]
CONTACTS
CONSORTIUM