Summary of Projects



Projects I have worked on in reverse chronological order with a brief description of my main responsibilities.



  • MTEVAL (2004-2006) (Quality models and resources for the evaluation of machine translation). Funded by Swiss National Science Foundation. The main objective of this project is to develop a web-based resource (FEMTI) dedicated to context-based evaluation of machine translation software. Main Responsibilities: the critical analysis of evaluation metrics for machine translation; surveying current real-world users and usage of machine translation.


  • PARMENIDES (2003-2005) (Ontology Driven Temporal Text Mining on Organisational Data for Extracting Temporal Valid Knowledge IST-2001-39023). A project to develop a text mining application for three organisational users. Main responsibilities: local project leader; developing a framework for evaluating the Parmenides system (including building a quality model and defining metrics for sub-components in information extraction, data mining, ontology development), eliciting user requirements, liaising with users and developers to facilitate the application of the framework.  (Described in: Underwood & Lisowska, (2006))


  • TQPRO (2000-2002) (Translation Quality for Professionals IST-1999-11407) A project to improve existing translation technology for professional and end-user use. Main responsibilities: project manager; responsible for the development of the translatability checker, a tool to assess the suitability of a text for translation using an MT system, (see Underwood & Jongejan (2001) ) experimental research into improving the quality of MT output by including lexical semantic information.


  • EUROMAP/HOPE (2000-2002) (HLT Opportunity Promotion in Europe IST-1999-12595) a dissemination project to boost market take-up of the results of national and European HLT RTD projects. Main responsibilities: dissemination and promotion of language technology in Denmark including maintaining the website, supporting our European partners in producing case studies and articles on language technology in their countries.


  • TRANSROUTER (1998-2000) (Translation Router LE-8345). A project to build a decision support system for translation managers to decide how best to carry out a translation project. Main responsibilities: local project manager; responsible for co-ordinating the design and development of component tools and profiles across the whole project and, at CST: the development of the repetitiveness detector and version comparison tool and developing profiles for translation jobs and backend systems.  (see Underwood & Jongejan (1999))


  • LCC (1999) (Language Consulting Centre MLIS-16).  LCC provided an on-line language technology consulting and advice service to small or medium sized companies wishing to introduce or make the best use of language technology products. Main responsibilities: producing a case study on the introduction of language technology in a localisation company 


  • GRASP (1999) An internal Danish grammar development project to build large grammatical resources which can be used in a number of applications. Main responsibilities: investigating the feasibility of re-using and building upon the existing resources available; in particular the proper treatment of immediate dominance and linear precedence in Danish.


  • EAGLES-II (1997-1999) (Expert Advisory Group on Language Engineering Standards LE3-4244). A follow-up project to consolidate and disseminate the evaluation framework developed in the EAGLES Evaluation Working Group. Main Responsibilities: design and maintenance of the web site and mailing list; planning and contributing to the new Handbook of Current Practices (Case studies on Translation Memories and Grammar Checkers); liaising with outside contributors to the handbook.


  • ELRA (1996-1997) Contract from European Language Resources Association Distribution Agency (ELDA) for the development of a draft validation manual for electronic lexica. Main responsibilities: project manager; development of a the draft manual and contributing to its sister document on standards for creating lexica. (see Underwood (1998) )


  • PILOTPROJEKT: Undersøgelse af udvalgte teksttypers egnethed til maskinoversættelse (1997). A pilot project for Lingtech (a machine translation customer) to assess the feasibility of translating new text types with their existing Machine Translation system. Main responsibilities: defining evaluation criteria and specifying methods for measuring them. (see Povlsen et al (1998) )


  • LINDA (1994-1996) (Linguistic Specifications for Danish: MLAP 93-09).  The project focused on developing specifications for core linguistic phenomena in Danish, most notably: inflectional morphology, phrase structure and constituent ordering, and predicate argument structure in terms of typed feature structures (akin to HPSG). The final product of the project is an integrated manual for Danish covering these core phenomena.   Main responsibilities: project manager; development of specifications for inflectional morphology, coordination and agreement in Danish. Overall responsibility for the scientific quality and coherence of all the workpackages and the editing and production of a full scale manual derived from these results. The target users of the manual are computational linguists seeking a solid theoretical and practical basis for their implementations of Danish.  The manual (339 pages) can be obtained in hardcopy from  Center for Sprogteknologi


  • TEMAA (1994-1996) (A Testbed Study of Evaluation Methodologies: Authoring Aids) Main responsibilities: development of measures for assessing the coverage and usability of spelling checkers; the derivation of weightings; application of the evaluation model to grammar checkers.


  • RELATOR (1994) (European Network of Repositories for Linguistic Resources: LRE 62-056) Main responsibilities: completing a survey of grammatical resources within the EU and particularly the Nordic countries.


  • COBALT (1992-1994) (Construction Augmentation and Use of Knowledge Bases from Natural Language Documents: LRE 61-011).  COBALT was a text categorisation project whose application was a message routing system for newswires. Its aim was to incorporate linguistic and semantic processing to improve accuracy and recall. Main responsibilities: Collaboration on the development of a robust text analysis module which can analyse ill-formed input with particular emphasis on the treatment of unknown words. Responsible for developing the interface between the output of the parser and the semantic analysis component..


  • PLUS (1991-1992) (A Pragmatics-Based Language Understanding System: ESPRIT P5254) The aim of PLUS was to build a robust natural language dialogue system as an interface to a variety of databases. Main responsibilities: The design of the interface between the natural language and knowledge base components of the system and experimental work on inferencing using abductive reasoning and conceptual modelling.


  • EUROTRA (1985-1992) (Multilingual Machine translation project). Main responsibilities: the development and implementation of large scale grammars for the English analysis module in a multilingual machine translation system.  Within the larger team I had sole responsibility for the analysis and implementation of certain linguistic phenomena (e.g. relative clauses, control structures, tense and aspect, adverbial phrases, subordinate clauses). As a basis for this work I carried out independent linguistic research on English in order to determine appropriate representations for the different linguistic phenomena, as well as some contrastive linguistic research. Latterly I was responsible for the supervision of junior staff in lexical coding work, and, in collaboration with colleagues, I carried out a comparative evaluation of DBMSs for dictionaries.