STILUS Trans

A software library for word-by-word translation of text between different languages

It is based on dictionaries of unidirectional translation (there is one per each couple of languages and translation direction) containing information of translation of forms. Besides, in the languages supporting them, they include the inflection of gender, number and case (for example, for Spanish, it translates singular/plural words and masculine/feminine words).

The unidirectional and bidirectional modules of translation between Spanish and all the co-official (vernacular) languages from Spain (Catalan/Valencian, Basque language and Galician language), as well as English, French, German, Italian and Arabic, are currently available:

• Spanish-Catalan-Spanish (es-ca, ca-es)
• Spanish-Basque language-Spanish (es-eu, eu-es)
• Spanish-Galician language-Spanish (es-ga, ga-es)

• Spanish-English-Spanish (es-en, en-es)
• Spanish-French-Spanish (es-fr, fr-es)
• Spanish-German-Spanish (es-de, de-es)
• Spanish-Italian-Spanish (es-it, it-es)
• Spanish-Arabic-Spanish (es-ar, ar-es)

Besides, the Spanish-Portuguese, Spanish-Polish, Spanish-Romanian, Spanish-Hebrew, Spanish-Russian and Spanish-Turkish as well as cross-language dictionaries between other languages than Spanish are being developed.

The dictionaries include the usual forms of each source language, in the categories of nouns, adjectives, verbs, adverbs, determinants, prepositions and conjunctions without being the semantic field limited (it covers general vocabulary used in this language, including terms from the colloquial language). In particular, it also includes more numerous proper nouns having a direct translation (for example, "Arantxa" into "Arancha" or "Irunea"/"Iruña" into "Pamplona"). Besides, for each of them, there is a plaintext dictionary, easily modifiable, in which translations can be included manually.

The main STILUS Trans characteristics are:

• Simplicity: it offers a very simple API, with a function that receives as an entry a chain with the text to be translated, the language in which it is written (it does not incorporate the automatic detection of languages) and the target language, and, as a result, it gives the word-by-word translation of this text into the specified language, in a understandable format easy to post-process.

• Coverage: availability of the unidirectional and bidirectional modules of translation between Spanish and more and more other languages. Logically the size and quality of the dictionaries are variable for each couple of languages but they are constantly checked and improved.

• Versatility: The STILUS Trans translation engine is distributed as a static library or a dynamics code, for Windows or Linux, to be integrated directly into C/C++, Visual Basic, Java and other languages using native methods. 

• Efficiency: The base of the translation engine is the TrieLib© library (library for management of Trie structures, developed by Daedalus and the Universidad Politécnica de Madrid).

• Maintenance: The STILUS Trans engine and translation resources are periodically updated. The modules of couples of languages have been developed all along the last four years as part of the participation of Daedalus in the CLEF European forum on multilingual information retrieval (http://www.clef-campaign.org).