William Shakespeare all the works that we associate with his person? In the mid-nineteenth century, they began to emerge
doubts about the author of the works attributed to the English playwright. A stream of academics did not quite believe that everything that was said that the Bard of Avon had written actually came from his pen. Their conspiracy theories, more or less successful, could have been resolved outside debates and opinions if they had had artificial intelligence tools a couple of centuries ago.
Petr Plechc, a researcher at the Institute of Czech Literature, has developed a machine learning system that will delight these skeptics by determining what parts of the play
Henry VIII they were really written by him and to what extent his contemporary and successor at the head of the theater company The King's Men participated in his creation.
Plechc enter their model from four works by each author:
Coriolano and Cimbelino, written by Shakespeare;
Monsieur Thomas and
The price of women, of the pen of
John fletcher. By analyzing its content, the Czech researcher's machine learning system extracted a thousand words and frequent rhythmic cadences to determine the style of each author. Once trained, the system was tested with
Henry VIII, where he could apply what he learned to determine which fragments were written by one author or another.
Method used by Plechc validates the hypothesis of the British author James Spedding, the first to say – in the nineteenth century – that Henry VIII could not be attributed exclusively to Shakespeare. Then, he stated that he had noticed evidence that certain scenes denoted a style that led to thinking that the work had been written in collaboration with Fletcher. He even proposed a separation of the scenes that he considered each one had written that substantially coincides with the results produced by the Czech algorithm. Of course, the level of precision used by the algorithm is considerably higher.
"The method of progressive attribution suggests that the particular scenes are mainly the work of a single author and that their contributions correspond approximately to what Spedding proposed",
the researcher maintains to conclude the report of his initiative. "Faced with the attribution of Spedding, our results show that the same scene could be written by both authors."
The distribution at the time of writing the different scenes may not be canonical for a reason. According to the algorithm, the author sometimes changes at the end of certain scenes, which may serve to achieve greater agreement when moving from one act to another.