Transcript Document

Teaching and Training in
Human Language Technology
Frank Van Eynde
University of Leuven
Part I - Flanders
•
•
•
•
•
•
•
Walter Daelemans (Antwerp)
Véronique Hoste (Ghent)
Jean-Pierre Martens (Ghent)
Frank Van Eynde (Leuven)
Hugo Van hamme (Leuven)
Astrid van Wieringen (Leuven)
Werner Verhelst (Brussels)
Overview of part I
1.
2.
3.
4.
5.
6.
7.
8.
structure of higher education programs
the four associations
Antwerp
Brussels
Ghent
Leuven
inter-university programs
conclusions
1. Structure
• Bachelor (180 study points)
- professional bachelors
- academic bachelors
• Master (min. 60 study points)
- initial masters
- advanced masters
2. The four associations
• Decision of Flemish government to
organize higher education in four
associations
• Each association includes at least one
university and one or more institutes for
higher education
• Relevant for HLT are especially the
universities and the institutes for
translators and interpreters
3. Antwerp
• University of Antwerp
–
–
–
–
7 Faculties distributed over three campuses
app. 10,000 students
23 Bachelor programs
86 Master and 40 Advanced Master programs
• Institutes for higher education
– Karel De Grote, Artesis Hogeschool Antwerpen,
Plantijn Hogeschool, Nautical Sciences School
– Total of app. 20,000 students
Faculty of Arts
• BA Language and Literature (180 SP)
– Two specialization choices from French, Italian,
Spanish, English, Dutch, Theatre-Film-Literature,
(History, Philosophy)
– Brief introduction to Computational Linguistics (for all
students): 15 hours
• MA Linguistics (60 SP)
– Specialization Computational Psycholinguistics
• Corpus Linguistics, Computational Linguistics, Computational
Psycholinguistics, Computational Models of Language
Understanding
• 3 to 5 students per year
• will probably become a 120 SP program
Karel de Grote
• School for Translation and Interpretation
– Starting up CL research in cooperation with
universities
• Subtitling, readability checking
– Will become a department within the
University of Antwerp
– Limited CL-related teaching
• translation tools
4. Brussels
• Vrije Universiteit Brussel
– 8 Faculties
– 9.200 students
– 28 Bachelor programs
– 60 Master programs
• Erasmus Hogeschool Brussel
– 4.200 students
Faculty of Arts
• Bachelor of Linguistics and Literature (180 SP)
• choice of two languages out of Dutch, French,
English, Spanish, German and Italian
– Working knowledge of all; in depth for two
• Master of Linguistics and Literature (60 SP)
– Option profiles (either 12 or 18 SP versions)
• 1 or 2 languages
• literature, linguistics, psychology of language or language
pathology
• Latin
Faculty of Engineering
• Master in Engineering Science: electronics
and information technology (120 SP)
– Profile Multimedia (24 SP)
• Digital Speech and Audio Processing (6 SP)
Erasmus Hogeschool Brussel
• Language and Media (Danish, English,
French, German, Italian, Spanish)
– Ba: Applied Linguistics (180 SP)
– Ma 1: Interpreters (Tolken)
– Ma 2: Translators (Vertalen)
5. Ghent
• Ghent University
– 11 faculties
– app. 30000 students
• Members of Association
– Hogeschool Gent
– Artevelde-hogeschool
– Hogeschool West-Vlaanderen
– app. 25000 students
Ghent University
• Master in Electrical Engineering
– elective course on speech processing (4 SP)
5 students / year
• Master in Computer Science
– course on speech processing (4 SP)
20 students / year
• Bachelor in Speech Therapy
– course on alternative communication through speech
& language technology (6 SP)
40 students / year
Hogeschool Gent
• Bachelor of translation studies
Ba3: Introduction to Language Technology (200 stud.)
• Master of translation studies (10 stud.)
Computatonial Linguistics
• Postgraduate Dutch and Translation (30 stud.)
– Introduction to Language Technology
– Computer-aided Translation
• Doctoral school (10 to 15 stud.)
– Introduction to Language Technology
6. Leuven
• Katholieke Universiteit Leuven
- 14 faculties
- app. 34.000 students
• 13 associated institutes
- distributed over 23 towns
- app. 51.000 students
- Applied linguistics: Lessius (Antwerp) and
HUB (Brussels) app. 1.100 students
Faculty of Arts - Bachelor
• Bachelor in language and literature (180 SP)
• choice of two languages out of Dutch, English,
German, French, Spanish, Italian, Latin and
ancient Greek
• BA3 contains a module of eight study points in
which students specialise in either literature or
linguistics or classical culture
• the linguistics module contains a.o.
computational linguistics (4 SP) and corpus
linguistics (4 SP)
• the bachelor paper (9 SP)
Faculty of Arts - Master
• Master in linguistics (60 SP)
• Common core (18-24 SP)
• Three options (12-18 SP)
-Computational and variational linguistics
-Diachronic and comparative linguistics
-Linguistics and language teaching
• Master thesis (18 SP)
Faculty of Engineering
• Master in Electrical Engineering: ICT (120 SP)
– Dutch programme: 3 options
• integrated electronics
– Digital signal processing: advanced topics (6 SP)
– Masters thesis (24 SP)
• Telecommunications
– Digital signal processing: advanced topics (6 SP)
– Masters thesis (24 SP)
• Multimedia
–
–
–
–
Speech and audio processing (6 SP)
Digital signal processing: advanced topics (6 SP)
Design assignment multimedia (6 SP)
Masters thesis (24 SP)
– English programme:
• Digital signal processing: advanced topics (6 SP)
• Design assignment multimedia (6 SP)
• Masters thesis (24 SP)
Faculty of Medicine
• Bachelor in Speech therapy and Audiology
Sciences (180 SP)
• Master in Speech therapy and Audiology
Sciences (60 SP)
• The program focusses on speech and
hearing
Interfaculty program - I
• Advanced Master in Artificial Intelligence (60 SP)
• Faculties of Science, Engineering, Arts,
Philosophy, Psychology and Medicine
• Common core
• Three options
-Engineering & Computer Science
-Cognitive Science
-Speech & Language Technology
Interfaculty program - II
Speech and Language Technology
• Linguistic Theories & Artificial Intelligence
• Phonetics & Speech Synthesis
• Speech Recognition
• Natural Language Processing
• Language Engineering Applications
• Text-based Information Retrieval
• Internship in an SLT company or research
institute (12 weeks, 20 SP)
Interfaculty program - III
• since 2001 58 students have completed
the Speech & Language Technology
program (an average of 7 to 8 per year)
The European Master in Language
and Speech
• Network of fourteen European universities which
offer a program in SLT that meets commonly
defined requirements about content and format
of the program
• Edinburgh, Sheffield, Saarbrücken, Bonn,
Stuttgart, Erlangen, Barcelona, Leuven, Utrecht,
Brno, Aalborg, Lausanne, Athens and Patras
• Annual Master School (one week in July)
• certified by ISCA and EACL
Lessius
7. Interuniversity program - I
• Master of Advanced Studies in Linguistics (60
SP)
– Leuven, Brussels, Antwerp, Ghent
– 2 students per year
• Background courses
– CL-related:
• Computational Tools for Linguistics (Antwerp)
• Corpus Linguistics (Leuven)
Interuniversity program - II
• Specialization options
– Comparative Linguistics (Ghent)
– Interdisciplinary Linguistics (Antwerp)
• Computational Linguistics
– Cognitive Artificial Intelligence, Automatic Text Understanding,
Linguistic Theories and AI, Intensive course / apprenticeship,
thesis
– Multilingual foreign language learning (Brussels)
– Cognitive and Functional Linguistics (Leuven)
• Will be phased out as soon as the two-year MA
in Linguistics is accepted
8. Conclusions - I
• language technology is mainly taught at
Faculties of Arts (esp. Antwerp and
Leuven), at institutes for translating and in
the interuniversitary Master in linguistics
• speech technology is mainly taught at the
Faculties of Engineering in Leuven, Ghent
and Brussels
• combining them is possible in the
advanced Master in AI program (Leuven)
Conclusions - II
• the number of students who follow courses
in language and speech technology is too
low to justify the creation of special Master
programs, but high enough to survive as
an option within broader existing programs
• the number of qualified staff is sufficient to
cover a wide range of HLT topics, both at
introductory and advanced levels
Part II – the Netherlands
•
•
•
•
•
•
•
John Nerbonne (Groningen)
Arjan van Hessen (Twente)
Antal van den Bosch (Tilburg)
Helmer Strik (Nijmegen)
Maarten de Rijke (Amsterdam)
Michael Moortgat (Utrecht)
Elisabeth d’Hallewijn (NTU)
Overview of part II
1.
2.
3.
4.
5.
6.
University of Amsterdam
University of Groningen
Radboud University Nijmegen
University of Utrecht
University of Tilburg
University of Twente
1. University of Amsterdam
• Multiple programs with HLT components,
– embedded in the Faculty of Science or
Humanities
– level: BA and MA
• Programs
– Science: BA computer science, artificial
intelligence, information science have up to
30 EC worth of HLT; MA computer science,
artificial intelligence, information science,
logic 1 or 2 years, with up to 60 EC on HLT
University of Amsterdam
• Programs (cont’d)
–…
– Humanities: BA in taalwetenschap with
selected modules in HLT, philosophy with
some HLT ingredients; similarly for MA in
general linguistics (one or two years) and
philosophy (philosophy of language)
2. Groningen University
•
Embedded in Faculty of Arts, collaboration with AI and Computing Science
(Fac. of Mathematics & Natural Sciences)
• level: BA and MA Information Science; MA Erasmus Mundus Language and
Communication Technology (LCT) (with Saarbrucken, Malta, Prague, Metz
en Bolzano)
• Courses and credits
* BA: NLP (20), programming incl. projects (30-40), statistics (10), logic (10), DB
(10), IR (10), Web incl. XML (15), information systems (10)
* MA: NLP (10), Machine Learning (10), Sem. Web (10), Computer-Mediated
Communication (10), Thesis (20), optional internship instead of 1 course (10)
* Erasmus Mundus LCT: see MA courses plus Seminar in Statistical Technology
(10), small research project (10 a 20), handwriting recognition (10),
discourse & pragmatics (10)
• Number of students: BA Information Science: 180/HLT: 50 - MA Information
Science: 60/HLT: 15 - MA LCT: 120/Groningen: 10
• Staff: 4,5 fte, occasionally postdocs and temporary lecturers
3. Radboud University (Nijmegen)
• Embedded in Faculty of Arts
• BA Linguistics: some HLT courses (all 5 ECTS)
• Minor Language & Speech Technology (L&ST)
20 ECTS, i.e. 4 x 5 ECTS courses
• Master Language & Speech Technology (L&ST)
60 ECTS, 5 & 10 ECTS courses
• 2 year research master Language and
Communication (L&C) together with Tilburg
120 ECTS
4. University of Utrecht
Cognitive Artificial Intelligence
• Participating faculties: Faculty of Humanities (Arts/Philosophy), Computer Science,
Social Sciences (Psyconomy)
• Level: BA, MA (2 year)
• Courses (all 7.5 ECTS credits):
* BA: Linguistics for Cognitive Artificial Intelligence, Logic Programming, Natural
Language Processing; Computational Syntax; Computational Semantics; Semantic
Variation
* MA: Foundations Language/Speech Technology; Semantic Web; Linguistic Resources;
Language and Learning ; Logical Methods in NLP; Conceptual Semantics
• number of students: BA: 60, MA: 15.
• Staff: 6
Linguistics
Courses: see above
• Introduction in HLT is compulsory for BA Linguistics
• The 1 year HLT Masters program will be suspended due to a lack of interested
humanities students
5. University of Tilburg
•
•
•
•
•
•
•
HLT is embedded in the program Communication and Information Sciences, Faculty of the
Humanities
BA: some HLT-courses;
MA: HLT-oriented Master-track "Human Aspects of Information Technology" (60 SP), in English
MA: Research Master Language & Communication in collaboration with Nijmegen (120 SP)
target group of MA: graduated BA students CIW, linguistics, information sciences, computer
science, information management, knowledge technology; premaster HAIT aims at graduated
college students computer science.
•
•
Courses:
Focus on language technology and information retrieval. core courses: Natural Language
Processing; Information Search, Retrieval, and Recommendation; Dialogue and Interface; Text
Mining; Knowledge Representation and Knowledge Management.
1 year Premaster HAIT 2009: also "Games and AI", "Cognitive Modeling".
Research Master: Language and Speech Technology (Advanced), Pragmatics.
•
•
Number of students: Total CIW BA: 150/, CIW MA: 150. MA track HAIT: 15
Staff: in 2009: 7.
6. University of Twente
•
•
•
•
•
•
•
HLT at the Human Media Interaction-group (HMI) is part of the Department
of Electrical Engineering, Mathematics and Computer Science
level: BA/MA
target group: students computer science
Courses & credits: UT is a technical (oriented) university with a clear focus
on applicability. The focus is on spoken and written man-machine dialogues
and unlocking of written and spoken information (= archives).
Number of students: HLT-courses: 30-60 students; graduations at HMI: ca
25/year
Staff: 8 (mainly part-time at the HMI-group)
In addition training for companies, organizations and institutions, f.e. master
classes for call center managers, data and communications specialists,
police and emergency services etc.