Automatic Classification of Research Documents using Textual Entailment

Abstract

Exploring the accumulative nature of Internet documents has become a rising issue that requires systematic ways to construct what we need from what we have. Manual and semi-manual document classification techniques have facilitated retrieval and maintenance of document repositories for easy access; however, they are customarily painstaking and labor-intensive. Herein, we propose a document classification model using automatic access of natural language meaning. The model is made up of application, business, and storage layers. The business layer, as a core component, automatically extracts sentences containing keywords from research documents and classifies them using the geometrical similarity of their sentential entailments.