An Automatic Partitioning of Gutenberg.org Texts

Abstract

Over the last 10 years, the automatic partitioning of texts has raised the interest of the community. The automatic identification of parts of texts can provide a faster and easier access to textual analysis. We introduce here an exploratory work for multi-part book identification. In an early attempt, we focus on Gutenberg.org which is one of the projects that has received the largest public support in recent years. The purpose of this article is to present a preliminary system that automatically classifies parts of texts into 35 semantic categories. An accuracy of more than 93% on the test set was achieved. We are planning to extend this effort to other repositories in the future. 2012 ACM Subject Classification Computing methodologies; Computing methodologies → Language resources