Workshop: Implementing Terminology
29 November 2019, 10:00 - 17:00
Hilton Brussels Grand Place, Carrefour de l'Europe, Brussels
Topic of the Workshop
Terminology is omnipresent in environments dealing with specific domains and with large amounts of documentation, such as DSIs and their national stakeholders. Understanding terminology is very important in various scenarios, such as authoring, translation and searching databases. In multilingual scenarios, selecting the proper translation for a term is an additional challenge. Because of the importance of terminology, language technology tools, such as translation memories (TM) and machine translation (MT) systems, need to be terminology aware. This workshop deals with the role of terminological resources in such tools.
The workshop has three sub-topics:
- To discuss the manual or automatic construction of terminological resources and concept systems (ontologies).
- To discuss the exploitation of these resources in language technology tools.
- To illustrate the creation and use of terminological resources in international organizations.
|10:00-10:10||Welcome (Luc Meertens, CrossLang)|
|10:10-10:30||Objectives of DG Connect (Ainars Freimanis, DG Connect, European Commission)|
|10:30-11:15||Building a termbase (Hendrik Kockaert, Hamad Bin Khalifa University & KU Leuven)|
|11:30-12:00||Extracting Term Candidates (Els Lefever, University of Ghent)|
|12:00-12:30||eTranslation TermBank (Andrejs Vasiljevs, Tilde)|
|12:30-13:00||Terminology in Translation Memories (Iulianna van der Lek, Ciudin, KU Leuven) |
|15:45-16:15||Terminology in IATE and the new possibilities of its integration with translation and terminology tools (Manuel Leal, Council of the EU)|
|16:15-16:45||Terminology and Concept Systems at WIPO (Cristina Valentini, WIPO)|
|16:45||End of the Workshop|
Information about the talks
Objectives of DG Connect (presentation)
The overall objective is to contribute to the development of CEF Automated Translation as a “multilingualism enabler” for CEF DSIs, online services linked to CEF DSIs, and other relevant public online services. By consequence, DG Connect also aims at collecting and distributing multilingual terminological resources and making use of such resources in the eTranslation system.
Building a termbase (presentation)
This presentation describes the properties of terminology databases, discusses term metadata (such as definitions and linguistic information), concept systems (specifying the relations between terms), multilingual termbases, exchange formats, and the sharing of resources by terminologists and translators using the cloud. The problem of aggregating terminological data is discussed, involving the filtering of duplicates and inconsistencies (quality assurance).
Extracting term candidates (presentation)
While terminology databases can be constructed manually, various tools exist that support the terminologist or translator in the detection of terms in documents: by extracting a list of term candidates, these tools allow for populating termbases semi-automatically. This presentation describes tools that make use of linguistic and statistical procedures in order to extract monolingual or bilingual term candidates.
eTranslation TermBank (presentation)
The eTranslation TermBank action aims at providing new terminological resources to the language resource pipeline of CEF eTranslation for three sector-specific domains, i.e. health, business legislation and customer protection, in order to support three CEF DSIs (eHealth, e-Justice and Online Dispute Resolution). These terminological resources cover all official languages of the EU plus Norwegian and Icelandic.
Terminology in translation memories (presentation)
In environments dealing with specific domains, translators need to be able to identify terms in the source text and may be required to use predefined translations only. In this presentation, functionalities of TMs for identifying terms, proposing the proper translation, and finding terminological inconsistencies are discussed.
Terminology in machine translation (presentations: 7a. Terminology in NMT-Crosslang, 7b. SDL Machine Translation and Terminology.pdf, 7c. SMART Workshop Terminology.pdf, 7d. SYSTRAN Implementing Terminology)
When a terminology database is available, an MT system should take its contents into account and should translate terms consistently. There is no guarantee that this is the case in a generic MT system: for instance, terms may be translated using a translation equivalent from another domain. This presentation discusses terminology related strategies in statistical MT and discusses the problem of making neural MT systems terminology aware, for instance for morphologically rich languages.
Terminology at IATE (presentation)
This presentation gives an overview of the importance of terminology work at the EU institutions and the practical aspects of it.
In that respect the following topics will be tackled:
- Interinstitutional collaboration for terminology and the new IATE project – the process of development of the new IATE and accompanying projects (cleaning, consolidation, normative collections, project management)
- The new IATE – new features enabling integration with CAT tools and other tools
- Short live demo of IATE, namely the different retrieval/export possibilities
- Future developments
Terminology and Concept Systems at WIPO (presentation)
Terminology resources are currently used in WIPO Translate SMT and NMT systems as training data. In turn, SMT and NMT systems allow to add equivalents to existing entries, including WIPO Pearl, in which MT equivalents are offered as “candidate translation proposals”. Terminology systems are hooked up in translation management systems. Concept systems are developed and are made accessible via the Concept Map Search in WIPO Pearl and a combined keyword function exploiting related concepts for document retrieval purposes is offered to users.
- No labels