The Sentence-Aligned European Patent Corpus
Wolfgang Täger · 2011
This paper describes the creation and the content of the Sentence-Aligned Euro-pean Patent Corpus. The corpus contains more than 130 million sentence pairs for 6 European languages. With more than 76 million sentence pairs, to our knowledge, the EN-DE sub corpus is the largest bilingual sentence-aligned corpus. For other language pairs, work has started to obtain sub corpora of simi-lar size. The error rate of sentence align-ment was very low even in the absence of language specific resources. 1 Translations in the patenting process Patent protection is granted by Patent Offices with temporal and territorial limitations. Each Patent Office has a specific language regime al-lowing patents to be filed in one or several lan-guages. To obtain broad geographic coverage, it is therefore necessary to translate patent applica-tions and to file these translated patents with oth-er Patent Offices. This may be the case for claim-ing priority, for entering a national or regional phase in the PCT procedure, or in Europe for