I will explain my thoughts and approach statistically to NLP, such as Statistical Machine Translation (SMT) in another post. The goal is constructing parallel corpus with limited resources. I will build a trilingual corpora for English, Thai, and Vietnamese concurrently to teach my Chat-bot.[1]
I don’t have trilingual translators who are fluent in all three sources and target languages so I have to use the next best thing. I am fluent in spoken Vietnamese and English and my wife is fluent in Thai and English. It will take a lot of time to collect such examples and I am hoping that my clients will help me train my bots.