Extract terminology · feed it into the process · QA against the termbase · ISO 17100 & ISO 18587
Terminology Database
The Governing Layer for Consistent Specialised and AI-assisted Translations
Trados · ECQA · LICS · SDL · Quanos · tekom · ISO 9001
A specialised translation is only as consistent as the terminology it is built on. Glossary, terminology and terminology management describe three levels of maturity of the same concept: a glossary is the entry point — an agreed list of terms. When structured, multilingual and equipped with definitions, contexts and sources, it becomes a terminology database (termbase). When continuously maintained and governed, it becomes terminology management.
With machine translation and AI, this foundation becomes more important. While AI systems can formulate fluently, they are blind to terminology, choosing plausible rather than binding designations and switching them without prompting. Only terminology that has been fed in steers the output towards reliability. There are two key processes: extracting terminology and using terminology.
Terminology only works once it has been built systematically and applied consistently. Three steps turn word lists into a governing resource:
Extract
We obtain candidate terms through term mining from your existing documents, past translations and translation memories, and then have them reviewed by subject experts and approved by you.
Feed in and use
The approved terminology is integrated into the translation tools. In the CAT tool and the MT engine, it enforces your naming conventions. An automated check verifies the output against the termbase before post-editing in accordance with ISO 18587.
Maintain and manage
A termbase is a living resource. New terms, corrections and approvals are continuously maintained with clear roles, ensuring consistent and traceable data.
Request a project & receive expert advice
Do you have technical documents that need translating? Send us 1–2 sample files. You will receive a specific recommendation and a quote within 24 hours.
The terminology work sounds like a big project. Getting started isn’t. We meet you where you are today and, step by step, turn your existing material into a structured, AI-ready termbase.
Start from where you are
A simple Excel list or an existing glossary is enough; no major project is needed.
We extract, rather than demand
Through term mining, we obtain candidate terms from your existing documents, glossaries and translation memories. We then define the permitted and non-permitted spellings for each term. You don't have to supply everything yourself.
Translate and enrich
The terms are transferred into the target languages and supplemented with definitions, context and usage notes. This results in one unambiguous designation per language.
Validate and approve
Entries are reviewed by subject experts and approved by you. A term only becomes binding after approval. This is how a structured, multilingual termbase emerges, with permitted and non-permitted terms, definitions, and context.
Visible and usable
Through the browser-based tool, everyone involved can easily view and maintain the terminology online.
Maintained and fed into AI
New terms are added, outdated ones are phased out The approved terminology governs MT and CAT tools, providing immediate benefits such as consistency and fewer correction loops.
Terminology quick start: We convert your existing material into an AI-ready starter termbase, providing an easy entry point into ongoing terminology management. terminology management.
How termbase, glossaries and translation memories work together
The glossary, termbase and translation memory are often confused However, they serve three different purposes and only realise their full value together. full value when they work together.
Glossary: the entry point. It is an accessible list of key technical terms with a short explanation, intended for lookup and shared understanding. See what such a collection looks like in our public glossary.
Termbase: the binding control layer. It is structured, multilingual and binding, and for each term it provides the permitted and non-permitted designations, as well as a definition and context. The termbase determines which designation is used in each language in CAT tools and AI.
Translation Memory: the memory. It stores entire verified sentence pairs and reuses them for similar texts, ensuring consistency at sentence level and reducing the effort required for updates. Find out more on the page about our translation memory systems.
These three layers need people to govern them in order to work together. Our project managers and language operators connect language, tools and processes. They maintain the termbase and glossary, keep the translation memory clean and ensure that the AI uses the correct terminology. This way, responsibility for quality remains with humans throughout.
When is a termbase worthwhile?
A termbase is worthwhile when...
- … translations recur or grow: documentation, updates, product variants.
- … multiple languages, stakeholders, or suppliers come together and consistency needs to be organized.
- … content is safety- or liability-relevant and an incorrect term can be costly.
- … AI and machine translation need to be governed rather than operating freely.
- … products or systems have a long lifespan and terms must remain stable for years.
A lean solution is sufficient when …
- … it is a one-off, small translation with no follow-up projects.
- … only one language direction is involved and there is little specialist terminology.
- … tone of voice matters more than fixed terminology, for example in marketing and communication texts.
- … you’re just getting started. In that case a simple glossary is often enough as an entry point and your termbase can grow from it later on.
Your terminology belongs to you
Ihre Fachbegriffe bleiben Ihre Fachbegriffe. Wir legen für jeden Kunden abgegrenzte, firmenspezifische Bestände an: eigene Termbank, eigenes Translation Memory, eigene Daten, strikt getrennt von anderen Kunden. Ihre Texte und Begriffe werden ausschließlich für Ihre Übersetzungen genutzt, nicht zum Training von Übersetzungs-Engines. Ihre Terminologie stellen wir Ihnen auf Anfrage in offenen Standardformaten (TBX) bereit, ohne Abhängigkeit von einem einzelnen Werkzeug. Das Translation Memory entsteht darüber hinaus aus unserer Übersetzungs- und Prüfleistung; dessen Bereitstellung vereinbaren wir im Einzelfall mit Ihnen.
This is backed by clear processes and technology: GDPR-compliant processing, servers in the EU and protected, continuously maintained IT systems. Non-disclosure agreements are standard practice. Read more about this under IT Security and Data Protection.
Terminology online: all in one place
The termbase and glossary are not stored as files on a drive, but in a browser-based tool which is accessible without installation. Stakeholders from technical, translation and editorial departments search for terms across all languages, view definitions and context, propose new designations and approve them. This keeps the terminology current in one place instead of it going stale in scattered lists.
Related pages
Do you want to do more than just understand terminology? Do you want to build it and put it to use? These pages take you further.
Frequently asked questions
Do I actually need a termbase, or is a glossary enough?
How do you build a termbase for us?
Can we reuse existing material?
Does the terminology belong to us, and can we take it with us?
How does terminology influence our AI-assisted and machine translation (MT) processes?
How many languages can a termbase cover?
How is terminology kept up to date?
Your terminology is in good hands
Translations you can rely on
Do you want to build your terminology or make it AI-ready? Send us your existing material. We'll recommend the right approach and provide a quote.
ISO 17100 · ISO 18587 · ISO 9001: audited processes, clearly assigned data
