Thursday, December 10, 2015

URSP Student Vankhanh Dinh Researches Lojban as an Interlingual in Machine Translation


I've always been fascinated by the limits of what a computer can do, like whether a computer can understand human language. To it, all human languages are foreign, so how can it receive this information and interpret the meaning behind it? This field, called Natural Language Processing, is focuses on developing computer tools for human languages with pragmatic goals like providing cheap translations between two languages.  Natural language services such as Google Translate, Bing Translate, and Babel Translate have served countless individuals in translating emails, documents and even websites-- thus bridging the gap between people. When I was very young, my family could have benefited from such tools as my parents did not speak English well and carried pocket dictionaries at all times in case they needed to translate something quickly. In time, they were able to graduate from understanding the meaning of words to understanding the meaning of phrases, but here they were stuck. No book was sufficient to help them translate emails back and forth. When they found Babel Translate, they were able to work more productively in a broader and wider world. Machine Translation had helped many immigrant families like mine and it serves as an important and useful tool for diplomats, tourists, and researchers. Unfortunately, Natural Language Processing is technically and algorithmically difficult, thus translation is still imperfect and remains an active area of research in computer science. In contrast to natural languages, humans have also designed some languages from scratch. These are often called constructed or artificial languages and are typically designed with a special purpose. The best known constructed language, Esperanto, has the goal of being an easy-to-learn, peace-promoting language. I recently gained some familiarity with a relatively new constructed language called Lojban whose design philosophy was to minimize ambiguity.  This makes Lojban analogous to a computer programming language for humans. I realized that it is possible that, if Lojban were used in machine translation of natural human languages, the translations produced between two languages, like English and Vietnamese, might be significantly improved. Improved translation tools would have broad impacts for people globally. Through this effort of studying machine translation of human languages I hope to gain a deeper understanding of natural language processing and doing research.