Eighteen years since the founding of Google and ten years since the launch of Google Translate, machine translation of texts in multiple language pairs on Google Translate has significantly improved and become more accurate. Oded Katiti is a computer scientist with a particular interest in artificial intelligence. He has also taken a keen interest in deep learning for machine translation.
Ever since computers entered the market and became available to the masses, scientists have been looking into the possibility of using its logistical power to break down language barriers. The reason behind Google Translate success in that time is based almost entirely on its unique access to vast quantities of translated text, sentences, and words that can be searched for likely matches in seconds. Most of the data was in fact taken from transcripts of United Nations meetings, where human translators and interpreters translate into six languages, and later included data of the European Parliament, translated into 23 different languages. Next, Google incorporated texts from its book-scanning projects and other internet sources, and added even more data to the syntactical database.
Today Google Translate offers an expedient, accurate service, and uses machine learning techniques to auto-translate. As Oded Kariti explains in his ARTIC collection, artificial intelligence allows Google Translate to learn in much more complex ways, and now thanks’ to this new type of artificial intelligence called deep learning, which involves training neural networks on lots of data and then making predictions about new data, the system is capable of performing more difficult tasks. The previous system took sentences and broke them up into individual words and phrases without context to translate, whereas the new system is able to read and analyze sentences from beginning to end, as a cohesive collection of words. Furthermore, compared to the old system, the new system reduces translation errors in some languages by up to 60%. Through this improved machine translation, Google is one step closer to a global future in which Artificial Intelligence systems can recognize and respond to natural human language, quick and efficient. It is fascinating to see how a research development is being transferred to industry and adopted in a product, says Educause member Oded Kariti.
We just sent you an email. Please click the link in the email to confirm your subscription!
OKSubscriptions powered by Strikingly