Developments in Irish Language Technology - Dr.Aodhán Mac Cormaic Director of Irish Department of Culture, Heritage and the Gaeltacht Government ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Developments
in
Irish
Language
Technology
Dr. Aodhán
Mac
Cormaic
Director
of
Irish
Department
of
Culture,
Heritage
and
the
Gaeltacht
Government
of
Ireland
Developments
in
Irish
Language
Technology
Achoimre / Summary
• Current status of Irish technology.
• Details of work already underway.
• Research demonstrates that Irish, like some major
world languages, is falling behind English in the Digital
Age.
• Government efforts to tackle this problem:
Ø Digital Plan for the Irish LanguageInvestment
by
Department
of
Culture,
Heritage
and
the
Gaeltacht
€9m invested
in
Irish
language
digital
and
technology
sector
since
2006.
Over
€1m
p.a.
over
last
3
years.
Projects
Funded:
• Abair.ie
– voice
synthesis
• Dúchas.ie
– Bailiúchán
na
Scol
1938
/
Schools
Collection
1938
• Ainm.ie
– biographical
details
of
people
involved
in
Irish
language
activities
• Logainm.ie
– place-‐names
online
• Royal
Irish
Academy
– historical
Irish
language
dictionary
• TechSpace
as
Gaeilge
• Dublin
City
University
-‐ MOOC
in
Irish
and
Irish
Traditional
CultureInvestment by Department of Culture, Heritage and the Gaeltacht (cont.) Irish Language Terminology for the EU Terminology Database (IATE) • Annual grant of €231,000 to Dublin City University. • Irish is now the 13th largest of the languages in the database and the largest of the new languages! • Over 72,000 terms translated into Irish do date. • Important work due to strategy to end derogation by 2022.
Investment by Department of Culture, Heritage and the Gaeltacht (cont.) www.gaois.ie • Search engine on www.gaois.ie site allowing searches for legal texts. • 9m words in this corpus, half in Irish and half in English. • Developed by Dublin City University
Investment by Department of Culture, Heritage and the Gaeltacht (cont.) Tapadóir: Machine Translation System • DCU research – statistic-‐based. • Trinity College Dublin – rule-‐based. • Hybrid system combining both.
Tapadóir
Tapadóir
European
Language
Resource
Coordination
(ELRC)
• EU
Commission
supporting
public-‐sector
bodies
across
Europe
to
provide
multi-‐lingual
public
services.
• Machine
Translation
system
to
be
developed
as
part
of
Connecting
Europe
Facility
(CEF).
• 2nd Phase
of
data
collection
process
currently
underway.Investment by Foras na Gaeilge Online Dictionaries • Irish version of Foclóir Béarla-‐Gaeilge, 90% of which is complete, available on www.focloir.ie. • Three other dictionaries – Béarla/Gaeilge (1959), An Foclóir Beag (aon teangach) agus Foclóir Gaeilge-‐ Béarla (1978) – all available on www.teanglann.ie.
2012 META NET Report:
The Irish Language in the Digital Age
Language
Processing:
level of
support
for
language
technology
for
30
European
languages
Excellent
Support Good
Support Reasonable
Intermittent
Poor
or
No
Support Support Support
Béarla
(English) Gearmáinis Bascais Íoslainnis
Iodáilis Bulgáiris Cróitis
Fionlainnis Danmhairgis Laitvis
Fraincis Eastóinis Liotuáinis
Ollainnis Gailísis Máltais
Portaingéilis Gréigis Rómáinis
Spáinnis Gaeilge
Seicis Catalóinis
Ioruais
Polainnis
Sualainnis
Seirbis
Slóvaicis
Slóivéinis
Ungáiris2012 META NET Report:
The Irish Language in the Digital Age
Machine
Translation:
level of
support
for
language
technology
for
30
European
languages
Tacaíocht
den
Tacaíocht
mhaith Tacaíocht
réasúnta Tacaíocht
bhriste Tacaíocht
lag
nó
scoth gan
tacaíocht
Béarla
(English) Fraincis Gearmáinis Bascais
Spáinnis Iodáilis Bulgáiris
Catalóinis Danmhairgis
Ollainnis Eastóinis
Polainnis Fionlainnis
Rómáinis Gailísis
Ungáiris Gréigis
Gaeilge
Íoslainnis
Cróitis
Laitvis
Liotuáinis
Máltais
Ioruais
Portaingéilis
Sualainnis
Seirbis
Slóvaicis
Slóivéinis
Seicis2012
META
NET
Report:
The
Irish
Language
in
the
Digital
Age
Text
Analysis:
level of
support
for
language
technology
for
30
European
languages
Tacaíocht
den
Tacaíocht
mhaith Tacaíocht
Tacaíocht
bhriste Tacaíocht
lag
nó
scoth réasúnta gan
tacaíocht
Béarla
(English) Gearmáinis Bascais Eastóinis
Fraincis Bulgáiris Gaeilge
Iodáilis Danmhairgis Íoslainnis
Ollainnis Fionnlainnis Cróitis
Spáinnis Gailísis Laitvis
Gréigis Liotuáinis
Catalóinis Máltais
Ioruais Seirbis
Polainnis
Portaingéilis
Rómáinis
Sualainnis
Slóvaicis
Slóivéinis
Seicis
Ungáiris2012 META NET Report:
The Irish Language in the Digital Age
Speech
and
Text
Resources:
level of
support
for
language
technology
for
30
European
languages
Tacaíocht
den
Tacaíocht
mhaith Tacaíocht
Tacaíocht
bhriste Tacaíocht
lag
nó
scoth réasúnta gan
tacaícoht
Béarla
(English) Gearmáinis Bascais Gaeilge
Fraincis Bulgáiris Íoslainnis
Iodáilis Danmhairgis Cróitis
Ollainnis Eastóinis Laitvis
Polainnis Fionlainnis Liotuáinis
Sualainnis Gailísis Máltais
Spáinnis Gréigis
Seicis Catalóinis
Ungáiris Cróitis
Ioruais
Portaingéilis
Rómáinis
Seirbis
Slóvaicis
SlóivéinisOur New Approach: A Digital Plan for the Irish Language • Long-‐term plan required in order to improve technologies in various sectors. • Expert team from DCU and Trinity College. • Planto be published late 2017.
Terms
of
Reference
for
Working
Group
To
prepare
cost
estimates
for
a
10
year
action
plan,
taking
into
account
the
following:
• that
the
annual
funding
which
will
be
available
to
implement
the
Plan
will
be
based
on
the
financial
resources
available
in
2015,
with
the
expectation
that
additional
resources
will
be
made
available
on
an
incremental
annual
basis:
• that
additional
financial
resources
may
be
made
available
in
the
future
within
the
timeframe
of
operation
of
the
Plan
as
a
result
of
same
being
redirected
from
other
projects
being
funded
by
the
Department
of
Culture,
Heritage
and
the
Gaeltacht
at
the
present
time;
• that
the
ongoing financial
implications
for
the
State
and/or
the
private
sector
be
taken
into
account
when
making
recommendations
in
regard
to
expenditure;
and,
• that
the
Plan
will
recognize
those
items
the
development
of
which
cannot
be
achieved
during
its
timeframe
but
that
an
attempt
will
be
made
to
supply
cost
estimates
for
the
development
of
those
items.
Aims
of
the
Plan
• To
set
up
a
long-‐term
research
and
development
infrastructure
that
will
into
the
future
deliver
those
state-‐of-‐the-‐art
technologies
that
are
increasingly
vital
for
language
maintenance.
• In
the
plan,
the
basic
linguistic
and
phonetic
research
is
seen
as
providing
the
essential
resources
for
the
technology
development.
These
technologies
include
machine
translation,
text-‐to-‐speech
synthesis,
speech
recognition,
and
dialogue
systems
that
enable
speech-‐based
human
computer
interaction.
• These
core
technologies
will
enable
the
development
of
the
growing
number
of
applications
that
will
serve
the
Irish
speaking
public.
• These
technologies
are
particularly
vital
for
the
teaching/learning
of
Irish,
as
well
as
for
those
with
disabilities.
Contents of the Digital Plan? Digital documentation and linguistic analysis of the written and spoken dialects • Types of documentation work required include: phonetics, syntax, phonology, discourse and language acquisition. • This work is required prior to development of language tools.
Contents of the Digital Plan? Language Resources: Resources, Data and Knowledge Bases • Irish language corpora providing detailed information relating to the linguistic structure of the language. • These corpora are fundamental to all other research work.
Contents of the Digital Plan? Natural Language Processing (NLP) What are the processing tools required in order to process the data obtained from the corpora? • Examples: Morphological analysis, part-‐of-‐ speech tagging, parsing, semantic analysis, named entry recognition etc.
Contents of the Digital Plan? Natural Language Understanding (NLU) • The computer must be taught to understand the text before it rather than merely seeing it as a string of words • This technology is used in search engines such as Google, for example.
Contents of the Digital Plan? Speech Synthesis • Much work already carried out by TCD (abair.ie) • Review this work and lay out next steps
Contents of the Digital Plan? Speech Recognition: Conversion of spoken word to text • A Priority – already in use in major languages. • Examples: mobile phones, automated subtitling, translation of speech in one language to another language. • Challenging work due to dialects and young people’s voices, for example.
Contents of the Digital Plan? Machine Translation Systems • Tapadóir, ELRC, MT@EC. • Further research required.
Contents of the Digital Plan? Dialogue Systems • User inputs a query in text or speech, computer interprets the query and provides a response. • These systems combine speech synthesis and speech recognition technologies.
Contents of the Digital Plan? Information Retrieval • Systems required for accurate retrieval of material in Irish language (i.e. nuclear energy v atomic energy). • Google only has front-‐end in Irish!
Contents of the Digital Plan? Educational Applications • Computer-‐Assisted Language Learning (CALL) • Interactive games for schools and universities in Ireland and overseas. • Self-‐led learning etc.
Contents of the Digital Plan? Access for people with disabilities • Blind or partially-‐sighted • Speech difficulties • Learning difficulties (e.g. dyslexia)
Contents of the Digital Plan? Role of national and multi-‐national companies and of Government and the public • Private sector support essential for localisation of services. • Public support for crowd-‐sourcing initiatives (e.g. Meitheal Dúchas) provides added value. • State incentivising small companies to localise their services (websites etc.)
GaelTeic @ DCU • Government investment of €621,000 over 4 years • Areas of Natural Language Processing (NLP) research that require significant attention with respect to the Irish language. The areas have been selected to build rapidly upon existing work, and to achieve maximum impact and reusability of the tools, data and resources created. The work here builds upon the pioneering work to date and will serve to provide a firmer grounding on which downstream language applications can be built to support the use of Irish in a digital environment.
GaelTeic @ DCU • Broad range of objectives include furthering state-‐of-‐the-‐art syntactic parsing for Irish, augmenting syntactically annotated corpora, improving tools for understanding Irish social media text, establishing Irish as a language used in international shared task projects and ultimately training a new team of language technology experts with Irish language skills, a resource that is significantly lacking and considerably impacting progress in this field.
Next
Steps
• Publication
of
Plan
by
end
2017
• Ministerial
support
essential.
• Funding
• Review
and
update
every
5
years.
EndYou can also read
