Structural and Semantic Indexing for Supporting Creation of Multilingual Web Pages

Translating webpages by machine translation is the easiest and fastest way a webmaster can multilingualize his/her webpages. Machine translation, however, often causes unnatural and mistranslated sentences with meanings that webmasters do not intend. Therefore, we propose a method that helps the webmaster to create multilingual web pages while avoiding mistranslations by adding metadata about analyzed sentence structures and word meanings. We have developed a prototype system that implements our proposed method. We evaluate our system and prove that it is able to translate sentences that machine translation mistranslates.