Report 2003: Department of Computational Linguistics (ID)
Department of Computational Linguistics (ID): Report 2003
History
The Department of Computational Linguistics was established 1 Aug 1985 with the purpose of developing the academic basis for a computational linguistics programme, planning and teaching the courses of this programme, and doing research in computational linguistics, particularly within the following areas: Formal syntactic and semantic analysis of language for specific purposes (LSP), modelling and representation of knowledge relevant to LSP, natural-language interfaces, automatic translation, and computational terminology & lexicology. Within these areas, the main focus is on computational linguistic issues that involve Danish.
The department’s research staff are active in inter-departmental and inter-institutional research projects, and for a number of years the department has worked in close collaboration with the Danish Centre for Terminology (DANTERMcentret) on presenting the department’s research results in shared projects with a number of Danish companies.
The department is responsible for courses at graduate level in computational linguistics (CLM), at undergraduate level in IT and English, at professional master’s level in Master of Language Administration (MLA), and for single courses at the BA programme in business language. In addition, the department holds PhD courses, as well as graduate and undergraduate courses in the languages programmes, and Open University. The department issues a line of publications titled LAMBDA.
Key action areas and action plan for the department in 2004 and later
The department has become the leading centre for formal research in language (with main emphasis on Danish), where the results are used in NLP systems. We are now striving towards the following goals:
- The department must maintain this position
- The research in the department must meet international standards.
Criteria for success: At least in three core areas research must meet international standards, i.e. there must be international refereed publications, the department should host international conferences and network meetings, members of the department should be invited speakers at international conferences, and research results should be part of Scandinavian or international standards.
- The department must develop in close cooperation with the private and public sector.
Criteria for success: The department must be the preferred partner for companies, when it comes to consultancy, development of language technology and courses on language technology. At least two companies should chose to enter into projects such as joint corporate PhD projects, hosting students projects or research projects.
- The department offers study programmes and courses in language technology and computational linguistics at all levels and attracts new students.
Criteria for success: The programmes offered by the department must be the preferred choice of students, when it comes to language technology and computational linguistics. The number of students on the present programmes must increase. New programmes and courses must be developed.
- The department contributes to spreading the general knowledge of computational linguistics and language technology and attracting public focus and funding to the area.
Criteria for success: The department must be visible in the public debate on language technology through popular articles and focus in the media on language technology. The web pages of the department must be revised and supplemented with an online show-room for language technology. The different projects must focus more on the user interfaces to their programmes in order to make them more understandable and approachable in demonstration situations.
Outline of objectives for 2003
The general and specific objectives for 2003 are outlined below.
General objectives
The department wishes to enter into mutually binding collaborations with related institutions in the country (for instance, in the form of a consortium or a centre of excellence).
The department wishes to enter into mutually binding collaborations with related institutions in the country (for instance, in the form of a consortium or a centre of excellence).
The department wishes to consolidate and maintain the present profile and the present activities.
During 2003, the department will work on obtaining funding from external resources to create successors to ongoing projects.
During 2003, the department will work on obtaining funding from external resources to create successors to ongoing projects.
Areas of academic development
The development projects listed below were to be carried out, if the resources were sufficient for it:
The development projects listed below were to be carried out, if the resources were sufficient for it:
- To investigate the possibilities of establishing a Centre for Terminological Ontologies
- To put a larger effort into term extraction and parallel term extraction
- To work more on the use of XML and RDF applications in the area of linguistics
- To strengthen the collaboration within the department by local lectures and guest lecture series, research seminars and study groups.
Internationalisation
The internationalisation objectives includes:
The internationalisation objectives includes:
- having more members in the Advisory Board (five instead of three)
- developing the collaboration with the Advisory Board through meetings between the Board and the individual research groups, demonstrations of programmes, and mutual presentations
- extending the participation in international projects
- continuing the close contact with international researchers
- improving the English version of the department’s home page and include more publications
- assuming the secretariat of the international Association for Terminology and Knowledge Transfer, GTW (Gesellschaft für Terminolgie und Wissenstransfer).
Collaboration with the corporate sector
The collaboration objectives are:
The collaboration objectives are:
- to establish an alumni network for MLA and CLM graduates, in order to strengthen the collaboration with the business sector with respect to research as well as education
- to strive to make computational linguistics and language technology better known to the public and to Danish companies.
General objectives
Collaboration with other institutions:
The department has participated in a manifestation of interest for the establisment of a Center of Excellence for Information Navigation together with researchers from the Technical University of Denmark, Roskilde University and the University of Copenhagen. Although the manifestation did not lead to a call for an application, the participants have agreed to continue working on the establishment of a centre construction to support collaborative efforts between institutions.
The department has participated in a manifestation of interest for the establisment of a Center of Excellence for Information Navigation together with researchers from the Technical University of Denmark, Roskilde University and the University of Copenhagen. Although the manifestation did not lead to a call for an application, the participants have agreed to continue working on the establishment of a centre construction to support collaborative efforts between institutions.
Furthermore, the department has negotiated for a fusion with the Centre for Language Technology (CST), which was to change status from an independent institution to an integrated part of a university. The perspectives were very promising indeed, since a joint venture between the two institutions at CBS could have created the strongest position within computational linguistics in Scandinavia. Although there was good support from CBS to this project, CST finally chose to be placed at the University of Copenhagen.
A part of the obligations of CST in connection with the fusion process was to establish a national consortium of computational linguistics. CBS is one of the partners in this consortium, which also involves the University of Aalborg (AUC). During 2003 a number of joint efforts were made by the consortium to improve the situation for computational linguistics in Denmark. First and foremost the consortium participated in preparing a position paper on the state of the art in computational linguistics in Denmark for the Danish Research Council of the Humanities (Statens Humanistiske Forskningsråd, SHF). Next the consortium has collaborated on the establishment of a Danish National Text Archive (Dansk Sprogbank). Finally, a number of research applications are under preparation.
At present the formal basis of the consortium remains to be established.
At present the formal basis of the consortium remains to be established.
Consolidate and maintain the present profile and the present activities
Very good progress has been made in all activities
Very good progress has been made in all activities
Obtain funding from external resources to create successors to ongoing projects
The following applications were made during 2003:
The following applications were made during 2003:
- Statistical dependency based machine translation, DDK 9 mill (SHF with CST, granted: DDK 1,8 mill.)
- CMOL – center for basic research (SHF DDK 4,5 mill., granted DDK 4,5 mill.)
- Medi@tor - information navigation DDK 5,4 mill. (VTU with RUC and DTU, granted: 0)
- OntoQuery: Continuation of PhD activities DDK 2,5 mill. (STVF with RUC, SDU and DTU, granted: 0)
- Center for Terminological Ontologies – travel and network support DDK 90.000 (Hedorf, granted DDK 90.000)
- Via – Visual Interactive Analysis for the web DDK 150.000 (CBS Learning Lab with Bente Lihn Jensen FIRST/CBS).
Our overall impression is that the results are very satisfactory. Nevertheless, more effort should be made to secure funding for research into terminology and ontologies.
Establishing a Centre for Terminological Ontologies
The centre has been established with an internal funding of 265.000 DDK for three years by the department.
The centre has been established with an internal funding of 265.000 DDK for three years by the department.
Increase efforts form research on term extraction and parallel term extraction
The area is covered in teaching projects, PhD projects and student projects and is an important topic in most applications. Nevertheless, no separate research projects have been launched.
The area is covered in teaching projects, PhD projects and student projects and is an important topic in most applications. Nevertheless, no separate research projects have been launched.
XML and RDF applications in the area of linguistics
XML is part of the syllabus of the BA Language Technology and English. The work on the Tamino database granted by the company Software AG has resulted in a fully functioning application with a web server for teaching and research projects. An internal workshop on XHTML has been planned for spring 2004.
XML is part of the syllabus of the BA Language Technology and English. The work on the Tamino database granted by the company Software AG has resulted in a fully functioning application with a web server for teaching and research projects. An internal workshop on XHTML has been planned for spring 2004.
Strengthen the collaboration within the department by local lectures and guest lecture series, research seminars and study groups
The number of guest lectures and international seminars has increased.
The number of guest lectures and international seminars has increased.
More members in the Advisory Board (five instead of three)
The Advisory Board now consists of five members:
Prof. Klaus Dirk Schmitz (Fachhochschule Köln)
Prof. Barbara Hall Partee (Univ. Massachusetts)
Prof. Torbjørn Nordgård (Univ. Trondheim)
Prof. Hans Uzkoreit (new) (Univ. Saarbrücken)
Chief Developer Hans Dybkjær (new) (Prolog Development Center A/S).
The Advisory Board now consists of five members:
Prof. Klaus Dirk Schmitz (Fachhochschule Köln)
Prof. Barbara Hall Partee (Univ. Massachusetts)
Prof. Torbjørn Nordgård (Univ. Trondheim)
Prof. Hans Uzkoreit (new) (Univ. Saarbrücken)
Chief Developer Hans Dybkjær (new) (Prolog Development Center A/S).
Develop the collaboration with the Advisory Board
Four AB-members participated in our three days research seminar in November 2003. This was a very satisfactory and inspiring event, which gave a good point of departure for our further development. One member, who could not participate, nevertheless, visited the department the week before to discuss the progress made in different areas.
Four AB-members participated in our three days research seminar in November 2003. This was a very satisfactory and inspiring event, which gave a good point of departure for our further development. One member, who could not participate, nevertheless, visited the department the week before to discuss the progress made in different areas.
Extend the participation in international projects
The number of international projects has not increased in 2003.
The number of international projects has not increased in 2003.
Continue the close contact with international researchers
The department has become a member of ISO-comittees in the area of terminology. Yet, the participation in these meetings is very expensive and cannot be finances by internal sources. The OntoQuery project organised an international PhD school in May 2003 and has made contracts with international capacities for the continuation of these activites. The OntoQuery group also organised an international workshop in November 2003. The department has agreed to host TKE 2005, and the MT summit 2006.
The department has become a member of ISO-comittees in the area of terminology. Yet, the participation in these meetings is very expensive and cannot be finances by internal sources. The OntoQuery project organised an international PhD school in May 2003 and has made contracts with international capacities for the continuation of these activites. The OntoQuery group also organised an international workshop in November 2003. The department has agreed to host TKE 2005, and the MT summit 2006.
Improve the English version of the department’s home page and include more publications
There were no resources for this work due to a shortage of administrative staff, and it was left up to researchers and projects to maintain their own web pages. The department decided to await the design of the CBS corporate web site, which will be released in spring 2004.
There were no resources for this work due to a shortage of administrative staff, and it was left up to researchers and projects to maintain their own web pages. The department decided to await the design of the CBS corporate web site, which will be released in spring 2004.
Assume the secretariat of the international Association for Terminology and Knowledge Transfer, GTW (Gesellschaft für Terminolgie und Wissenstransfer)
Agreement has been made. Beginning in 2005.
Agreement has been made. Beginning in 2005.
Establish an alumni network for MLA and CLM graduates, in order to strengthen the collaboration with the business sector with respect to research as well as education
There were no ressources to start a formal alumni network. Nevertheless, cooperation has been strengthen with the alumi organisation Datalingvistik.dk and with the network of present and former MLA students. Thus, the department hosts the network meetings and evening seminars of Datalingvistik.dk. The informal contacts are used in connection with project applications. Thus, the companies NOVO Nordisk, and LingTech have agreed to provide texts and translations for the the Statistical Dependency-based Machine Translation project (see below).
There were no ressources to start a formal alumni network. Nevertheless, cooperation has been strengthen with the alumi organisation Datalingvistik.dk and with the network of present and former MLA students. Thus, the department hosts the network meetings and evening seminars of Datalingvistik.dk. The informal contacts are used in connection with project applications. Thus, the companies NOVO Nordisk, and LingTech have agreed to provide texts and translations for the the Statistical Dependency-based Machine Translation project (see below).
Strive to make computational linguistics and language technology better known to the public and to Danish companies
The department participates in ITEK, a network organised by the Confederation of Danish Industries (Dansk Industri), and has made more efforts to write more popular articles in magazines and newspapers. The department and the DANTERM centre contributed with talks to the Danish Language Technology Fair, and the department is represented in SHF’s consultancy group for Language Technology, where a position paper on language technology will be presented in the spring of 2004.
The department participates in ITEK, a network organised by the Confederation of Danish Industries (Dansk Industri), and has made more efforts to write more popular articles in magazines and newspapers. The department and the DANTERM centre contributed with talks to the Danish Language Technology Fair, and the department is represented in SHF’s consultancy group for Language Technology, where a position paper on language technology will be presented in the spring of 2004.
In the context of the MLA programme, an agreement was made with Medtronic A/S Denmark that provided text and translation material and agreed to act as a case company for the course on machine translation in the spring term.
In general, the department is very open to contacts with the private and public sectors, and quite a number of meetings were held in order to establish possible areas of cooperation. Together with the DANTERM centre the department has started collaboration with Pericon A/S, where Pericon has the role of identifying relevant corporate partners. This has led to an increase in the number of corporate contacts during 2003.
Below, the department’s research projects are described. Extended descriptions are given for new projects, whereas research reports from previous years should be consulted for full descriptions of ongoing projects.
EU projects
There were no ongoing EU projects in 2003. Nevertheless, applications for EU-fundings in 2004 were prepared.
There were no ongoing EU projects in 2003. Nevertheless, applications for EU-fundings in 2004 were prepared.
Projects with funding from other external resources
Pargram
The goal of the project is to analyse and describe a number of central linguistic phenomena in Danish and Norwegian within the linguistic theory of Lexical-Functional Grammar (LFG) and to implement the descriptions in a state-of-the-art grammar development environment. The goal is, by these means, to produce formal linguistic specifications that will expose differences and similarities between the two languages, and to develop computational linguistic resources for Danish and Norwegian that can be used in actual computational applications, as well as for research and teaching purposes.
The goal of the project is to analyse and describe a number of central linguistic phenomena in Danish and Norwegian within the linguistic theory of Lexical-Functional Grammar (LFG) and to implement the descriptions in a state-of-the-art grammar development environment. The goal is, by these means, to produce formal linguistic specifications that will expose differences and similarities between the two languages, and to develop computational linguistic resources for Danish and Norwegian that can be used in actual computational applications, as well as for research and teaching purposes.
OntoQuery
The purpose of the interdisciplinary research project Ontology-based Querying ( http://www.ontoquery.dk/ ) is the development of theories and methods for content-based information search in textual databases. Within the project, work is performed in parallel on the subjects of search, ontology, syntax/semantics, and prototype development.
The purpose of the interdisciplinary research project Ontology-based Querying ( http://www.ontoquery.dk/ ) is the development of theories and methods for content-based information search in textual databases. Within the project, work is performed in parallel on the subjects of search, ontology, syntax/semantics, and prototype development.
During 2003, particular work has been done on further development of the so-called ontoparser, on the basis of rules expressed in ontological grammar. The parser is implemented in the LKB system with the purpose of achieving analyses that include relations between concepts in the so-called Ontolog description language. Descriptions derived from the parser form the basis for evaluation of queries; in particular, the establishment of measures for closeness or “similarity” between descriptions, and “calculations” with these goals, have been in focus also in 2003.
During 2003, the project has presented papers at international conferences, and continues to participate in the European OntoWeb network ( http://ontoweb.aifb.uni-karlsruhe.de/
).
STO (Sprog Teknologisk Ordbase).
STO (Sprog Teknologisk Ordbase).
STO (Sprog Teknologisk Ordbase) is a project with the aim of developing a 50.000 word computational lexicon for Danish, including morphological, syntactic and to a certain extent semantic information, for use in the development of language technology applications. The project was initiated by the Danish Ministry of IT- and Research’s working party “IT in Danish”, which has set aside DDK eight mill. for the project for three years. The work is coordinated by the Centre for Language Technology, and will be completed by the end of February 2004.
SweDanes
This project deals with the comparative linguistics of Danish and Swedish speech, and is carried out in collaboration with the Department of Linguistics, University of Göteborg. The project is funded by NorFA.
The project is currently working on a book on Danish and Swedish spoken language with a clear data-driven point of view summing up the achievements of the past three years. The project finishes in 2004.
This project deals with the comparative linguistics of Danish and Swedish speech, and is carried out in collaboration with the Department of Linguistics, University of Göteborg. The project is funded by NorFA.
The project is currently working on a book on Danish and Swedish spoken language with a clear data-driven point of view summing up the achievements of the past three years. The project finishes in 2004.
NordTalk
This project deals with the establishment, exchange and utilisation of speech corpora. Representatives of all the Nordic countries, as well as Estonia, participate.
This project deals with the establishment, exchange and utilisation of speech corpora. Representatives of all the Nordic countries, as well as Estonia, participate.
Within the two NorFA projects NordTalk and SweDanes, Peter Juel Henrich has developed an algorithm for automatic (n-gram based) translation of spontaneous speech (in orthographic transcription). The method, which aims at application in speech technology, is corpus-based and uses as its basis two transcription corpora (both >250.000 words) in two different, but related languages. The output is a 1:1 translation of the most high frequency words (up to rank 300, which will include the function words in particular). Experiments have been performed with the large Swedish corpus Göteborg Corpus of Spontaneous Speech and the large Danish corpus BySoc. Preliminary results are promising: Among the 100 most frequent Swedish speech words, approximately 90% are translated correctly (for instance, måste -> skal) and the rest near-correctly.
Research projects in collaboration with The Danish Centre for Terminology
The overall objective of the Danish Centre for Terminology (DANTERMcentret) is to contribute to the development of Danish know-how within the fields of terminology and language technology, and to develop methods and tools for the creation and management of corporate term bases. Through the Centre, the department has improved its potential for establishing good and constructive collaborations with a number of Danish companies. The department has made an active contribution to the efforts to secure the continued existence of the centre, including participating in meetings and seminars and filing applications for funding of joint research projects.
The IT Terminology Project
The purpose of this project is to establish a database and a web page with advice on Danish IT terminology. The project is carried out in collaboration with the Danish Centre for Terminology, the Danish Language Council, the Institute of Computer Science at the University of Copenhagen, the IT University of Copenhagen, the Danish IT Industry Association a.o. The project has been funded by VILLUM KANN RASMUSSEN FONDEN (DKK 600.000 in 2003).
The purpose of this project is to establish a database and a web page with advice on Danish IT terminology. The project is carried out in collaboration with the Danish Centre for Terminology, the Danish Language Council, the Institute of Computer Science at the University of Copenhagen, the IT University of Copenhagen, the Danish IT Industry Association a.o. The project has been funded by VILLUM KANN RASMUSSEN FONDEN (DKK 600.000 in 2003).
Database for the Lykeion Thesaurus
This project deals with information about central concepts in relation to system analysis and construction. In 2002, the Centre for Terminology received a grant from the foundation Lillian og Dan Finks Fond (DKK 50.000), to initiate the project. In 2003 the following activities have been carried out: analysis, proposal for database structure and conversion of data.
This project deals with information about central concepts in relation to system analysis and construction. In 2002, the Centre for Terminology received a grant from the foundation Lillian og Dan Finks Fond (DKK 50.000), to initiate the project. In 2003 the following activities have been carried out: analysis, proposal for database structure and conversion of data.
ParaT
The project Parallel Texts covers the development of methods for automatically parallelising texts, and tools for the handling of parallel texts, establishment of parallel corpora and research into the use of parallel texts in computational linguistics, for instance in translation memory systems. Work on the further development of the corpus has been at a standstill, while the effort has been put into the development of a web interface for displaying parallel corpora on the internet, and a robot for the automatic collection of corpora.
The project Parallel Texts covers the development of methods for automatically parallelising texts, and tools for the handling of parallel texts, establishment of parallel corpora and research into the use of parallel texts in computational linguistics, for instance in translation memory systems. Work on the further development of the corpus has been at a standstill, while the effort has been put into the development of a web interface for displaying parallel corpora on the internet, and a robot for the automatic collection of corpora.
CMOL
The Center for Computational Modelling of Language (Peter Juel Henrichsen, Dan Hardt, Matthias Trautner Kromann, Peter Rossen Skadhauge) is funded by a special senior staff salary allocation. The purpose of the centre is to provide a framework for basic research projects with the aim of developing models for language processing, particularly from psycholinguistic and corpus-based criteria. The centre’s research is organised around three projects: “Grammar Acquisition”, with the aim of developing methods for automatic language acquisition from large corpora; “Discontinuous Parsing”, with the aim of developing methods for representation and processing within the theoretical framework of “Discontinous Grammar”; and “Interpretation”, with the purpose of developing semantic methods for representation and processing of linguistic phenomena both within and across sentences.
The Center for Computational Modelling of Language (Peter Juel Henrichsen, Dan Hardt, Matthias Trautner Kromann, Peter Rossen Skadhauge) is funded by a special senior staff salary allocation. The purpose of the centre is to provide a framework for basic research projects with the aim of developing models for language processing, particularly from psycholinguistic and corpus-based criteria. The centre’s research is organised around three projects: “Grammar Acquisition”, with the aim of developing methods for automatic language acquisition from large corpora; “Discontinuous Parsing”, with the aim of developing methods for representation and processing within the theoretical framework of “Discontinous Grammar”; and “Interpretation”, with the purpose of developing semantic methods for representation and processing of linguistic phenomena both within and across sentences.
CMOL is cooperating with the Indian Institute of Technology (IIT, Kanpur) and Benares Hindu University (Varanasi) about the development of speech technology in an Asian context. During 2004 CMOL will apply for EU funding in "The EU-Asia Information Technology and Communications Programme". CMOL has been working hard to raise external funding, as the internal funds were only given for two years. By the end of 2003 the project received a grant of 4.5 mill. DDK for three years from the Danish Reseach Council of the Humanities.
Centre for Terminological Ontologies (CTO)
The work of the Centre for Terminological Ontologies includes basic research as well as application oriented research and development. It is the purpose of CTO to work with methods, which can form the scientific basis for the development of ontology-related IT tools.
The work of the Centre for Terminological Ontologies includes basic research as well as application oriented research and development. It is the purpose of CTO to work with methods, which can form the scientific basis for the development of ontology-related IT tools.
CAOS
CAOS, or Computer-Aided Ontology Structuring (Bodil Nistrup Madsen, Hanne Erdman Thomsen and Carl Vikner, Bo Kranz Simonsen), aims at the development of a system for semiautomatic construction of concept systems by means of feature structures, on the basis of user-typed information. In 2003, the project has received DDK 126.640 from the department’s research allocation and DDK 90.000 from the foundation Hedorfs Fond. The funding has been spent on programming assistance, release and salary for emeritus professor Carl Vikner who is very active in the project.
CAOS, or Computer-Aided Ontology Structuring (Bodil Nistrup Madsen, Hanne Erdman Thomsen and Carl Vikner, Bo Kranz Simonsen), aims at the development of a system for semiautomatic construction of concept systems by means of feature structures, on the basis of user-typed information. In 2003, the project has received DDK 126.640 from the department’s research allocation and DDK 90.000 from the foundation Hedorfs Fond. The funding has been spent on programming assistance, release and salary for emeritus professor Carl Vikner who is very active in the project.
Formal terminological concept analysis
The work with CAOS has motivated the formulation of a new project: Formal terminological concept analysis. The purpose of this project is the development of theories and methods for formal description of terminological concept systems (ontologies). In terminological concept analysis it is insufficient to structure concept systems solely by the generic concept relation between superordinate and subordinate concepts. It is also necessary to use such concept relations as for instance the part-whole relation, the causal relation and the resultative relation. Therefore, it is necessary for the formal apparatus to be able to describe such polyrelational concept systems. The theoretical results obtained in the project will be tested and integrated in the CAOS system in the course of the development.
The work with CAOS has motivated the formulation of a new project: Formal terminological concept analysis. The purpose of this project is the development of theories and methods for formal description of terminological concept systems (ontologies). In terminological concept analysis it is insufficient to structure concept systems solely by the generic concept relation between superordinate and subordinate concepts. It is also necessary to use such concept relations as for instance the part-whole relation, the causal relation and the resultative relation. Therefore, it is necessary for the formal apparatus to be able to describe such polyrelational concept systems. The theoretical results obtained in the project will be tested and integrated in the CAOS system in the course of the development.
Web-ontologies
The aim of the project is to investigate and evaluate formalisms and tools designed to build web ontologies, i.e. ontologies, which are meant to support the use of mark up languages on the Internet.
The aim of the project is to investigate and evaluate formalisms and tools designed to build web ontologies, i.e. ontologies, which are meant to support the use of mark up languages on the Internet.
OntoQuery
The report on activities in the interdisciplinary research project Ontology-based Querying can be found in the section on projects with external funding.
The report on activities in the interdisciplinary research project Ontology-based Querying can be found in the section on projects with external funding.
Danish Grammar Checking Systems
The purpose of the project Danish Grammar Checking Systems is to develop reliable grammar checkers for a broad range of grammatical problems. A technique has been developed to deduce grammatical principles automatically by means of advanced machine-learning techniques, which are applied to syntactically annotated corpora.
The project continues within CMOL and has currently been extended with a writing assistant facility.
VIA
In 1998, the first prototype of the VIA programme was launched with funding from CTU (Center for teknologistøttet uddannelse), comprising a total of 1000 exercises for seven languages. The programme is used in linguistics courses in the IT and English programme, and at high-school level.
The purpose of the project Danish Grammar Checking Systems is to develop reliable grammar checkers for a broad range of grammatical problems. A technique has been developed to deduce grammatical principles automatically by means of advanced machine-learning techniques, which are applied to syntactically annotated corpora.
The project continues within CMOL and has currently been extended with a writing assistant facility.
VIA
In 1998, the first prototype of the VIA programme was launched with funding from CTU (Center for teknologistøttet uddannelse), comprising a total of 1000 exercises for seven languages. The programme is used in linguistics courses in the IT and English programme, and at high-school level.
In 2003, CBS Learning Lab granted DDK 265.000 for the adaptation of the programme to the web and the development of new excercise types and excercises for the Italian language. The project is led by Bente Lihn Jensen (FIRST, CBS) in cooperation with Sabine Kirchmeier-Andersen, Bo Krantz Simonsen, and Jacob Møller Christensen and CBS Learning Lab.
Multimedia in the Home
The goal of this project is to investigate the possibilities of automatically annotating information about music and making these as well as the corresponding recordings available for everybody who wants to listen to, download, acquire or just be informed about existing recordings of classical music. In the funding period, the project participants have been Mette Nelson, University of Southern Denmark, Kolding, and Steffen Leo Hansen of the department, who is in charge of the project.
The goal of this project is to investigate the possibilities of automatically annotating information about music and making these as well as the corresponding recordings available for everybody who wants to listen to, download, acquire or just be informed about existing recordings of classical music. In the funding period, the project participants have been Mette Nelson, University of Southern Denmark, Kolding, and Steffen Leo Hansen of the department, who is in charge of the project.
The Department’s Corpus Initiative
To ensure an optimal utilisation of the department’s IBM-sponsored servers, and to ensure the application of results from the ParaT project and others projects that involve corpora, the department’s staff work on establishing joint corpus resources at the Faculty of Languages, Communication and Cultural Studies. This work has been formalised in the Department’s Corpus Initiative.
To ensure an optimal utilisation of the department’s IBM-sponsored servers, and to ensure the application of results from the ParaT project and others projects that involve corpora, the department’s staff work on establishing joint corpus resources at the Faculty of Languages, Communication and Cultural Studies. This work has been formalised in the Department’s Corpus Initiative.
The department has been among the originators of the idea of establishing a Danish Language Bank (Dansk Sprogbank) for securing a standardised availability of Danish corpus resources and other resources for developing language technology. A preliminary project for the Language Bank has been drafted under the supervision of the Danish Standards Association and with support from the Ministry of Science and Technology.
Other research projects
Stanlex
Bodil Nistrup Madsen heads a project concerning classification and structuring of lexical data. The project was started in 1995 and has the purpose of developing a taxonomy and a description of the content of lexical data in databases and natural language systems, as well as models for the structural description of lexical data using SGML. Hanne Erdmann Thomsen joined the group in 2003.
Bodil Nistrup Madsen heads a project concerning classification and structuring of lexical data. The project was started in 1995 and has the purpose of developing a taxonomy and a description of the content of lexical data in databases and natural language systems, as well as models for the structural description of lexical data using SGML. Hanne Erdmann Thomsen joined the group in 2003.
Representation of transscribed speech
In collaboration with the NordTalk project, the Department of Linguistic, University of Göteborg, and the Department of Danish Dialectology, University of Copenhagen, Peter Juel Henrichsen is working on a tool for translation between the most widespread Scandinavian formats for representation and transcription of spontaneous speech – currently in particular the formats CorDiale, MSO06, BySoc and Danish Standard 2.
In collaboration with the NordTalk project, the Department of Linguistic, University of Göteborg, and the Department of Danish Dialectology, University of Copenhagen, Peter Juel Henrichsen is working on a tool for translation between the most widespread Scandinavian formats for representation and transcription of spontaneous speech – currently in particular the formats CorDiale, MSO06, BySoc and Danish Standard 2.
Sound transcription and syntactic tagging of the corpus “Spontan-tale” (Spontaneous speech).
The project, which is carried out in collaboration with the Department of Applied Linguistics at the University of Copenhagen (Nina Grønnum), comprises 20 speakers and approximately 25.000 running words. The tagging is carried out at CMOL and is to be funded by the Carlsberg Foundation in 2004.
The project, which is carried out in collaboration with the Department of Applied Linguistics at the University of Copenhagen (Nina Grønnum), comprises 20 speakers and approximately 25.000 running words. The tagging is carried out at CMOL and is to be funded by the Carlsberg Foundation in 2004.
The Danish Dependency Treebank (DDT)
The purpose of the project is to construct a Danish Dependency Treebank with 50-100.000 dependency annotated words, based on the Danish PAROLE corpus. A secondary purpose is that of composing a tagging manual for Danish, with information on the linguistic principles behind the treebank, including how a large number of linguistic constructions are analysed in Danish within Discontinous Grammar, the dependency formalism used by the project. The treebank should be available to general linguists for their corpus-based basic research, and to computational linguistics and companies for the development of language technology applications by means of feature-based machine learning methods. ( http://www.id.cbs.dk/~mtk/treebank )
The purpose of the project is to construct a Danish Dependency Treebank with 50-100.000 dependency annotated words, based on the Danish PAROLE corpus. A secondary purpose is that of composing a tagging manual for Danish, with information on the linguistic principles behind the treebank, including how a large number of linguistic constructions are analysed in Danish within Discontinous Grammar, the dependency formalism used by the project. The treebank should be available to general linguists for their corpus-based basic research, and to computational linguistics and companies for the development of language technology applications by means of feature-based machine learning methods. ( http://www.id.cbs.dk/~mtk/treebank )
Dependency annotation tool (DTAG)
The project is a natural continuation of M.T. Kroman’s former PhD project on Discountinous Grammar. 6000 lines of programming code has been written, making DTAG a fully usuable tool with respect to manual dependency annotation (a), display of dependency graphs (b), and syntactic search (c). This part of DTAG has formed the basis for the tagging of the first 60.000 words in DDT. Furthermore, algorithms have been developed for automatic parsing and lexicon building. Results can be seen at: http://disgram.sf.net/and http://www.id.cbs.dk/~mkt/dg .
The project is a natural continuation of M.T. Kroman’s former PhD project on Discountinous Grammar. 6000 lines of programming code has been written, making DTAG a fully usuable tool with respect to manual dependency annotation (a), display of dependency graphs (b), and syntactic search (c). This part of DTAG has formed the basis for the tagging of the first 60.000 words in DDT. Furthermore, algorithms have been developed for automatic parsing and lexicon building. Results can be seen at: http://disgram.sf.net/and http://www.id.cbs.dk/~mkt/dg .
PhD Projects
The department has four PhD students.
The department has four PhD students.
Tina Nielsen investigates computer assisted learning applications for teaching language technology. The purpose of the investigation is to formulate, discuss and test construction principles for computer assisted learning in relation to the teaching of LSP and language technology at university level. Tina Nielsen’s PhD project is financed by the Danish Research Council for the Humanities and affiliated to the Danish Centre for Terminology.
Nina Sværke Hansen (formerly Frederiksen) works on a contrastive study of the phenomena of ”extraction” in Danish and French, with the aim of implementation in Lexical-Functional Grammar (LFG). Nina Sværke Hansen is affiliated with the PARGRAM project. Nina has been on maternity leave since 29 May.
Ekaterina Mhaanna works on the formal properties of ontologies. She is affiliated with the OntoQuery project.
Lone Bo Sisseck investigates methods for identification and extraction of conceptual relations in Danish domain-specific texts, in order to be able to generate ontologies automatically. It is a very time consuming task to build an ontology manually. It will therefore be of great interest to be able to construct ontologies automatically or semi automatically.
By the end of 2003 Jakob Halskov was given a PhD grant by the Faculty to work on the detection of new words and terms in parallel corpora.
Other academic activities in 2003
The department’s staff are academically active in a large number of ways and have participated in many seminars, conferences and courses. Information about these activities is only included to the extent that they fall within the following five headings: PhD study programme, external activities, guest lectures, conferences and course development. Activities such as participation in CBS committees, course responsibility and coordination, supervision and evaluation of job applicants are not included.
PhD study programme
The department has participated in a joint application for the establishment of a Nordic Graduate School of Language Technology. An application was submitted to NORFA in June 2003. Funding has been granted in 2004. Dan Hardt is a member of the board, representing Denmark. http://ngslt.org/
The department has participated in a joint application for the establishment of a Nordic Graduate School of Language Technology. An application was submitted to NORFA in June 2003. Funding has been granted in 2004. Dan Hardt is a member of the board, representing Denmark. http://ngslt.org/
Network activities
The department is part of the computational linguistics network in Denmark, consisting of all computational linguistics environments. The purpose of the network is an exchange of research results and coordination of the computational linguistics research.
The department is part of the computational linguistics network in Denmark, consisting of all computational linguistics environments. The purpose of the network is an exchange of research results and coordination of the computational linguistics research.
International conferences
The department has hosted international and Scandinavian conferences and workshops in connection to the described projects.
Furthermore, GTW’s next international conference TKE 2005 (Terminology and Knowledge Engineering) is planned to be held in Copenhagen. See: http://gtw-org.uibk.ac.at/
Participation in collaborative projects
- Collaboration with Software AG about the XML database Tamino (sponsored in 2001)
- Preliminary discussions about joint research and education projects
- Collaboration with The Danish Centre for Dyslexia on the establishment of a sound based dictionary directed at the needs of dyslectic people
- Collaboration with the Danish Center for Terminology on business oriented projects .
Publications
The table below compares the number of publications in 2002 with 2003. The number of publications in 2003 is smaller than in 2002. In 2002 there were 31 publications, 56% in English. In 2003 there were 25 publications, 44 % in English. A possible explanation is the fact that the department in 2003 entered into a number of new research project, where publications have not yet reached the publishing phase.
Teaching acitivities
Teaching activities are described in detail by the programme directors of the various programmes in separate reports. This section gives an overview of the teaching activities of the department. It should be stressed that the department strives that all teaching activities are research based, since we consider teaching the best way to disseminate research results. By research based we thus mean courses, which in addition to leading theories in the field also present recent research results of the department, either by the researchers themselves or by a qualified teacher instructed by the researchers.
The Master’s programme in Computational Linguistics
In 2003 the department has published a qualifications profile for the master’s programme in computational linguistics. The profile documents the intended occupation of the candidates, the qualifications required to carry out this occupation and how the programme aims at providing the candidates with these qualifications. In addition the department has continued its work on implementing the new master’s programme in Computational Linguistics. The work on the master’s programme is documented in a separate report (“studielederberetningen”).
In 2003 the department has published a qualifications profile for the master’s programme in computational linguistics. The profile documents the intended occupation of the candidates, the qualifications required to carry out this occupation and how the programme aims at providing the candidates with these qualifications. In addition the department has continued its work on implementing the new master’s programme in Computational Linguistics. The work on the master’s programme is documented in a separate report (“studielederberetningen”).
MA supplementary course in Language Policy
The department has developed a course for students in the master’s programme in modern language and business communication. The content of the specialisation is the design and implementation of language policy in companies with a special emphasis on the IT tools available to support this work. The course pays special attention to localisation and is offered for the first time in autumn 2003.
The department has developed a course for students in the master’s programme in modern language and business communication. The content of the specialisation is the design and implementation of language policy in companies with a special emphasis on the IT tools available to support this work. The course pays special attention to localisation and is offered for the first time in autumn 2003.
The BA programme in Language Technology and English
A lot of work has been put into giving the programme its final form. The basis of this work was the experience gained from the programme IT and English, and a questionaire made by the evaluation unit. The results can be viewed at:
http://www.cbs.dk/studies/sprogtek/undervisning .
A lot of work has been put into giving the programme its final form. The basis of this work was the experience gained from the programme IT and English, and a questionaire made by the evaluation unit. The results can be viewed at:
http://www.cbs.dk/studies/sprogtek/undervisning .
BA in knowledge management and communication
In cooperation with the Center for Communication and the English Department, a new programme is being developed.
In cooperation with the Center for Communication and the English Department, a new programme is being developed.
The Master of Language Administration programme
In 2003 the MLA-programme was moved to the Sitescape learning space and has now only four seminar days per semester. All teaching activities are carried out in the learning space in small self-organising working groups. The MLA programme has been associated with CBS CELL (Center for executive learning and leadership).
Until august 2003 the programme had a specially assigned programme director, Ellinor Hansen, who made a special marketing effort by issuing a newsletter and established contacts to relevant companies.
In 2003 the MLA-programme was moved to the Sitescape learning space and has now only four seminar days per semester. All teaching activities are carried out in the learning space in small self-organising working groups. The MLA programme has been associated with CBS CELL (Center for executive learning and leadership).
Until august 2003 the programme had a specially assigned programme director, Ellinor Hansen, who made a special marketing effort by issuing a newsletter and established contacts to relevant companies.
Initiatives in further education
Carsten Hansen is currently developing courses on language technology for high school teachers. This was made possible by a grant from the faculty. The department is interested in further development in this area.
Last updated by Webmaster 28/11/2008
