Machine translation for names

High-quality machine translation is a work in progress for languages such as Arabic, Chinese, Russian, and Korean, but Rosette® has solved the very specific and ubiquitous problem of translating names. Human translators are excellent at translating sentences, but computers are superior when it comes to consistently applying complex transliteration standards to translating names.
Whether for gisting reports or translating entire name directories, Rosette’s name translator is fast and produces consistent spellings.

Accurate and smart

Rosette translates names using its knowledge of language-specific naming conventions. Our name translation tools are trained to recognize when to transliterate a name (based on spelling or sound), and when to translate the meaning (as with a personal title like “King”).
The name translator produces highly accurate, standardized English translations based on linguistic algorithms and statistical modeling. Rosette supports name translation in complex, non-Latin languages — such as Arabic, Chinese, Hebrew, Korean, Persian, Thai, and Urdu — and plugs into familiar document and spreadsheet applications.

Product highlights

• Supports multiple language pairs and writing systems
• Plugin for MS Word and MS Excel
• Fast and scalable
• Industrial-strength support
• Cloud, on-premise, and search plugin deployments

Spelling versus meaning

Words in a name can be translated according to their spelling/pronunciation or their meaning. Consider the following example:

Notice the title, “King,” and its placement. In any given language, a name may be native to that language or transliterated from another language. Consider محمود أحمدي نجاد (Mahmoud Ahmadinejad), whose name is native to Arabic, and جورج دبليو بوش (George W. Bush), whose name is not. In each case, Rosette attempts to provide the most logical translation. Thus from Arabic to English, Rosette offers “George Bush,” instead of the transliterated “Jurj Bush.”

Features of the name translator

• Consistently follows the user-chosen transliteration standard, such as the IC or U.S. Board on Geographic Names for translating from non-Latin scripts to English
• Distinguishes between when to generate “conventional spellings” of common or well-known names and when to transliterate “unrecognized” names; i.e., those not appearing in any known catalog of foreign names
• Automatically resolves name spelling ambiguities in the source language, such as partial vocalization of Arabic, or word segmentation in Chinese

Arabic Translation Sample

The following examples illustrate Rosette Name Translator’s ability to translate names between Arabic and English depending upon the name’s language of origin.

Sample output:

{
  "translation": "Mu'ammar Muhammad Abu-Minyar al-Qadhaf",
  "targetLanguage": "eng",
  "targetScript": "Latn",
  "targetScheme": "IC",
  "confidence": 0.06856099342585828
}