Terminology validation for MT output

Terminology validation for MT output Giorgio Bernardinello STAR Group Wiesholz 35, 8262 Ramsen Switzerland giorgio.bernardinello@star-group.net Abstract WebTerm Connector is a plugin for STAR MT Translate which combines machine translation with validated termi- nology information. The aim is to pro- vide “understandable” information in the target language using corporate language and terminology. 1 WebTerm WebTerm is STAR’s web-based terminology management system that can be seamlessly inte- grated into STAR MT Translate in order to search for terms and display terminology infor- mation. 2 STAR MT Translate STAR MT Translate is STAR’s web-based ap- plication for MT systems that can also be inte- grated into Microsoft Office products. STAR MT engines are based on customer-specific transla- tions and terminology. As additional feature, the web-based application can access the company’s Transit translation memory. 3 WebTerm Connector The Web Term Connector plugin allows users to search for translations of terms in the company’s dictionaries within the STAR MT Translate environment. Furthermore, it is possible to display additional terminology information for the source and the target language terms, if this is available in the dictionaries. © 2018 The authors. This article is licensed under a Creative Commons 3.0 licence, no derivative works, at- tribution, CC-BY-ND. 3.1 Searching for terms On the assumption that the company dictionaries contain validated terminology, the translation is searched for in the dictionaries first. If no translation is found, especially for longer strings and sentences, the TM is then checked for perfect matches. If none are found, the MT engine will provide the translation. 3.2 Information about terms If more information on parts of the source or the target sentence is needed, the user can simply highlight the text to search for in the company- specific dictionaries. If the word is found, its translation and the available terminology information are displayed at the bottom of the same web page. The customer can define what information from the dictionaries should be shown. The major advantage of this solution is that users can quickly access company terminology without having to switch between browser tabs. 3.3 Translation of terminology information Sometimes the dictionaries contain definitions or context descriptions in just in one language, e.g. the target language. But the user might need them in the source language in order to under- stand them, which is why MT technology is also integrated into the terminology area. This makes it possible to obtain an automatic translation of text parts of the dictionary simply by clicking the “Translate” button there. 3.4 Improved term searching functionality All strings that were highlighted for terminology searching but for which no match was found in the dictionaries are listed in a log file. The cus- tomer can then use this information to improve not only their WebTerm dictionaries but also their MT engines. P´erez-Ortiz, S´anchez-Mart´ınez, Espl`a-Gomis, Popovi´c, Rico, Martins, Van den Bogaert, Forcada (eds.) Proceedings of the 21st Annual Conference of the European Association for Machine Translation, p. 343 Alacant, Spain, May 2018.