How Beijerterm works
What's behind the search box — where the terms come from, how they're searched, and what's planned next.
What it is
Beijerterm is a curated, multilingual terminology database for technical translators, spanning Dutch, English, German, French and Spanish, with more languages added as new sources arrive. Every term pair carries full provenance: you can always see which glossary or reference work it came from, with author, year, languages, entry count and licence. It's free to search here, and built into Supervertaler as a SuperLookup resource.
How search works
Search runs on the server, not in your browser. When you type, only the matching results are sent back — the full dataset never leaves the server. Behind the scenes it uses SQLite's full-text search (FTS5), so it stays fast whether the database holds ten thousand entries or a million. You pick the source and target language with the From / To selectors (any available pair, in either direction), and results are grouped by headword, so every equivalent and synonym for a term shows together on one card.
Synonyms
Source glossaries often list several synonyms in one cell, separated by a delimiter such as a semicolon or a slash: for example windturbine; WEC → wind energy converter; wind turbine; WEC; WT; WTG. Beijerterm splits these apart, so each synonym becomes independently searchable and they surface as one another's equivalents. (Each source can declare its own separator — semicolon, slash, pipe, and so on.)
The sources
Everything is assembled from curated source files — published glossaries, term lists and reference works — each imported with its own column mapping and bibliographic metadata. Nothing is scraped or auto-aggregated. You can browse the full list, with provenance for each, on the Sources page. Term pairs are reorganised into a single curated database and served only as isolated, per-query snippets.
Under the hood
The site is a small static front-end; the database lives in Cloudflare D1 (edge SQLite) behind a lightweight API that returns only what each query needs. A private data core handles importing, normalising, de-duplicating and cleaning the source files before anything is published — so the live database is always a vetted snapshot of the curated master copy.
What's planned
Two things are scaffolded but not yet switched on:
- Concept grounding. Linking entries to open lexical resources such as Open English WordNet (and its Dutch counterpart) to power richer synonym grouping and "related term" suggestions.
- More open terminology. Selectively folding in openly licensed resources (for example the EU's IATE export) where the licence allows and the quality is worth it.
This page describes the current state honestly — when those features go live, it'll be updated to say so.