HLT Research and Development for Baltic Languages in Tilde

1
HLT Research and Development for Baltic Languages
in Tilde

• Andrejs Vasiljevs, Raivis Skadin
• Tilde

Riga, October 27, 2004
2
HLT needs of Baltic users

• Baltic users expect the same possibilities for
  their language as are available for other
  languages
• Raising number of users without English
  proficiency
• Bilingual/multilingual environment, language
  competition
• EU accession and globalization creates huge
  demand for translation

3
Prerequisites for HLT application development
HLT Industry
Scientific Research
4
Challenges for Baltic Languages

• Rich structure, inflectivity and complexity of
  languages
• Relatively small number of speakers, small market
• Few researchers active in HLT, insufficient
  scientific base
• Lack of specialists resulting from lack of
  targeted education programs

5
Tilde - Basic Facts

• Tilde leading Baltic developer of software and
  services that helps Baltic users to get most from
  the use of new technologies taking into account
  their language, business environment, culture and
  local information needs
• Established in 1991
• Offices in Riga, Vilnius and Tallinn
• 75 employees

6
Tilde - Areas of Activity

• Language and Reference Tools for Baltic Languages
• proofing tools, dictionaries, translation tools,
  fonts, search tools, speech tools, Encyclopedias
• Localization Services for ICT products
• translation and adaptation of interface and
  documentation for software and hardware products
• Information management solutions
• Streaming media

7
Localization Services Where Global Meets Local

• Requirement of users to have interface and
  documentation in Latvian, Lithuanian, Estonian
• Full cycle of localization services
• Intelligent translation using translation
  memory,e-dictionaries and other tools
• Professional team across the Baltics
• Global clients in local markets Microsoft, IBM,
  Ericsson, Nokia, Xerox, Hewlett Packard and
  others
• MS Windows, MS Office, Oracle, mobile phones
  examples of localized products

8
Proofing Tools

• More than 10 years of development
• Tools for Latvian and Lithuanian languages
• Spellchecker
• Intelligent Autocorrect
• Grammar checker
• Hyphenator
• Thesaurus
• Collaboration with Microsoft,developments for
  Lotus Notes

9
Demo Grammar Checker and Spellchecker
10
Demo - Thesaurus
11
Terminology Portal

• Partnership from NGO and private sector to
  provide common online database of terminology
  from all fields
• Initiated by LITTA(Latvian Information
  Technology and Telecommunication Association)
• Partners LITTA, LMT, Tilde, Terminology
  Commission
• Currently covers gt145 000 terms from 35 fields

12
(No Transcript)
13
Demo Text-to-Speech in MS Excel
14
Our Products

• Tildes Birojs 2002 - complete solution for
  working with computer in Latvian
• Tildes Biuras 2004 - complete solution for
  working with computer in Lithuanian
• gt 100 000 licensed users

15
Tildes Birojs and Tildes Biuras

• HLT Tools - Proofing tools, Dictionaries,
  Learning tools, Limited speech synthesis, Search
  facilities
• Language support for all popular operating
  systemsWindows 95/98/NT/2000/Me/XP, Windows CE,
  Linux, MacOS
• Text input tools - keyboard drivers and keyboard
  adjustment tools, fonts, converters
• OCR
• Document Templates
• Information resources smart tag references,
  eBooks and others

16
Web service Letonika.LV

• Encyclopedias (General, History, Poetry,
  Financial, Regional)
• Translation Dictionaries
• Terminology Dictionaries
• Catalog of Latvian Web
• Advanced search system (under development)

17
Search and information retrieval

• Established infrastructure for web crawling and
  indexing
• Search result ranking and filtering methods
• Advanced usage of Latvian/Lithuanian morphology
• Integrated with Latvian Web Catalog
• Wide coverage, constantly updated
• Work on style guessing

18
Demo Search in Latvian Web
19
(No Transcript)
20
Further Development

• Tildes goal
• To enable Baltic users to explore results of
  native language HLT in their everyday work
• Development directions
• Search and information retrieval technologies
• Machine Translation
• Speech technologies

21
Success factors

• Demand driven approach focusing on user needs
• Development of base technologies and integration
  in end-user products(like integration of
  morphology in dictionary)
• Iterative approach efficient delivery of first
  results and constant improvements

22
Challenges

• Further development in advanced areas (speech,
  machine translation, IR etc.) is very complicated
  and resource consuming
• Technology gap separating from large languages
  should be narrowed
• Baltic and EU cooperation is critical
• Researchers and industry developers should target
  EU programs with joint projects

23

• THANK YOU!

www.tilde.lv www.tilde.lt www.tilde.ee