We have created a story visualization system for literary works. A hint as to how this can be accomplished lies in language education; babies and small children grasp the meaning of long complex sentences by focusing on the keywords that they had already understood. The overall meaning of the story is then roughly reconstructed by the combination of these keywords . One of the problems that remain currently is how to reconstruct a story systematically from scattered keywords. A number of trials in robots and human interface machines have achieved a dialogue with humans by means of pattern recognition in the conversation. Such software development is now underway in artificial intelligence research, beginning with corpus classification, topic mining and language compilation. Another approach to analyzing a long story is to introduce advanced mathematics into the text data.
A second problem is how to present the meaning of the story to a third party. Different from the visualization for nature or physical phenomena, though which are often observed as beautiful image, the story itself includes emotional substance for the readers.
In the present work, the first attempt towards story visualization considers the two issues discussed above. Shakespeare’s plays have been selected as the target of the visualization since these are the most famous historical literary works and the structure of the sentences have been investigated in great detail. Although the results presented in this paper might seem primitive to literature researchers, we are confident that this initial step is necessary to pioneer future development. A possible achievement might be that someday a robot will enjoy reading a human letter, and will begin to write a literary work by the end of this century. The visualization of a story aimed at in this study, will hopefully become the basis of such a system that is realized in the future.
- Miyuki Yamada Graduate School of Letters, Hokkaido University, Sapporo, Japan
- Yuichi Murai Graduate School of Engineering, Hokkaido University, Sapporo, Japan
Full Text (PDF) p. 1141-1147