Secure Agents for Supporting Best-Balanced Multilingual Communication

There are technologies that support intercultural collaboration by allowing people to communicate more easily across the barriers of culture and language. However, sometimes user-sensitive information needs to be accessed. In best-balanced machine translation, a method that recommends the languages and machine translation services that should be used to assist multilingual group communication, user test scores must be disclosed to generate the language recommendations. There are various methods that can protect the data (test scores) and methods that allow simple statistic calculations, however, no existing method supports the complex calculations needed by the best-balanced machine translation method. This paper emphasizes the importance of user privacy in intercultural collaboration. We provide the initial idea and show how user test scores can be protected while supporting the recommendation system. We introduce a detailed example to discuss the design of a suitable user interface.