太田 彩 & 脇田 建: 小説内の動的人物相関図を用いた読書システム
Abstract: Books are reliable and excellent sources of information. However, reading requires long time constraints and the need to supplement the information from the text. Therefore, even if you have the desire and need to read, you may give up. For this reason, researches have been conducted to automatically extract information from novels and to assist the reader by representing the content of the story in a single page. However, there is a risk of spoiling. In this study, we propose a method to support reading by presenting a character correlation. In order to reduce the risk of spoilers, we propose a character correlation chart that changes as the reading progresses. We used two methods to add information: automatic extraction using natural language processing techniques and manual addition. In addition, by sharing the input information with others, we were able to reduce the time and effort required to input information. In order to verify the effectiveness of the proposed method, we constructed a reading support system using Aozora Bunko data. Finally, we discussed the degree of achievement and issues in comparison with the objectives.
Keywords: Visualization / Dynamic graph / Natural language processing