Artificial Intelligence helps to decipher ancient manuscripts

The ancient manuscript you see in the image below is called the Vonyich manuscript and it was written in the 15th century. It has troubled historians and cryptographers since it was discovered in the 19th century because no one could figure out the language it was written in. To help decipher the text, researchers turned to the AI community with computer science professor Greg Kondrak lending his expertise in natural language processing.

Kondrak and a grad student called Bradley Hauer set about trying to decipher the manuscript using computers for decoding the human language. The first step in the process was to figure out the language of origin which was written on delicate vellum pages with illustrations.

Using samples of 400 different languages from the "Universal Declaration of Human Rights" the researchers set about trying to identify the language. Originally the scientists thought it might be written in Arabic. After the duo ran it through their computer systems it was determined the language was actually Hebrew. The scientists said that it turned out over 80% of the words in the manuscript were in a Hebrew dictionary.

The team first turned to Hebrew scholars to decipher the text, but they were unsuccessful. The researchers then used Google Translate and it was able to come up with a sentence that was grammatical, and that people could interpret. That sentence was "she made recommendations to the priest, man of the house and me and people."

The researchers say that without historians of ancient Hebrew the full meaning of the manuscript may never be known. Kondrak says that he is looking forward to using the algorithms he and the student developed on other ancient manuscripts.

SOURCE: Ualberta.ca