Building semantically annotated corpus for text classification of Indian defence news articles

A large amount of textual data is generated online with rapid growth and technological advancement. Deriving interesting patterns like opinions, summaries and facts from the text data is a challenging task. Currently, there is no dataset for subjectivity/objectivity classification data in Indian National Security domain. A News dataset has been created for purpose of subjective/objective sentence classification. This paper defines the news corpus annotation guidelines and employs an inter-annotator agreement metric to assess the quality of the dataset. The proposed methodology also highlights different challenges and limitations of building a corpus in the National Security domain. The corpus can be utilized for research work in developing robust subjective/objective sentence classifier. Furthermore, text categorization experiments are conducted on corpus, demonstrates that neural network based classifier gives promising result. © 2021, Bharati Vidyapeeth's Institute of Computer Applications and Management.