Reading large volumes of texts is a rather fascinating but tedious task. Finding in a pile of mediocre stories something worthwhile is also not so simple, so the developers of Disney decided to shift this ungrateful work to the "shoulders" of artificial intelligence and created an algorithm capable of analyzing and evaluating texts.
Together with experts from the University of Massachusetts at Boston, the Disney Research team has created three neural networks that can work together to read the entire text and find out how good it is. One neural network deals with the analysis of individual parts of the stories, the second estimates how well all these parts are connected with each other, and the third one already looks at the full text and gives it its own estimate.
"Our neural networks are very good at analyzing the texts and predicting their possible popularity. You will not be able to use them in the jury of your writers contest, but in our future research they will definitely be useful, "says Albert Lee of Disney Research.
The latest development of Disney Research has shown good results, coping with the work by 18 percent better than its more primitive counterparts. Previously available algorithms at the disposal of scientists could not show such impressive results, and the new AI is predicted a great future. According to the developers, the algorithm gradually learns to understand the meaning of the works read by him, assesses their quality and, it seems, will soon learn to write good fairy-tale little books.
The article is based on materials .
Comments
Post a Comment