"Dictionaries" - a new raw data product
Leider ist diese Seite derzeit nicht in deutscher Sprache verfügbar.
Newsflash 5/2010
The EPO, in close cooperation with the EPC member states, has developed electronic machine translation dictionaries. Under the framework of this project, to date, the EPO had completed the following dictionaries:
- German - English; English - German
- Spanish - English; English - Spanish
- English - French; French - English
- English - Italian; Italian - English
- English - Portuguese; Portuguese - English
Future updates will be announced.
The electronic machine translation dictionaries are based on an Open Lexicon Interchange Format (OLIF)
The application is run on translation engines and the product has been custom-made for patent documents by using the IPC classification symbols to identify the technical context. The subfolders are organised by IPC subclasses.
The basis for the creation of the dictionaries and for training the translation engines, were the large collections of European patents that have been translated into the languages of the National Offices. The size of each of these corpora varies between 40.000 and 600.000 documents, depending on the existence of translated documents.
Below are some examples, extracted from the new OLIF structure and content manual, relating to different spellings of German words.
| Attribute | Description | Example |
|---|---|---|
| german-1 | Match vowels to stem | Schänke/Schenke |
| german-2 | "selbstständig" instead of "selbständig" | unselbstständig/unselbständig |
| german-3 | German spelling of non-German words | Soße/Sauce |
| german-4 | Write "f" instead of "ph" | Fantasie/Phantasie |
| german-5 | Write "r" instead of "rh" | Katarr/Katarrh |
| german-6 | Write "t" instead of "th" | Tunfisch/Thunfisch |
| german-7 | Write "zi" instead of "ti" | differenziell/differentiell |
| german-8 | Plural "ices" instead of "izes" | Indices/Indizes |
| german-9 | New spelling of non-German words | Campagne/Kampagne |
| german-10 | Repeat three letters without a hyphen | Schifffahrt/Schiff-Fahrt |
| german-11 | Write preposition and "weak" noun as two words | im Stande/imstande |
| german-12 | Write "nicht" in compound adjectives as a separate word | nicht öffentlich/nichtöffentlich |
| german-13 | Write "rein" in compound adjectives as a separate word | rein seiden/reinseiden |
| german-14 | Write "wohl" in compound adjectives as a separate word | wohl tuend/wohltuend |
| german-15 | Write non-German words with multiple parts as a single word | Bluejeans/Blue Jeans |
| german-16 | Write non-German words with multiple parts with a hyphen | Fall-out/Fallout |
| un | unspecified |
The European Patent Office Machine Translation Program (EMTP) uses a hybrid approach. In addition to the statistical method which makes use of the classifications allocated to each document, linguistic rules have also been applied, e.g. a rule for distinguishing the various parts of speech, i.e. nouns, adjectives and verbs.
- More information on the product
- Product manual