PIREN: Prediction of Intermediary Readers’ Emotion from News-Articles

Stimuli like text narratives from news articles and editorials trigger numerous emotions as a response in the readers. As seen in previous works of news documents classification, the attention has been centralized more towards writer’s perspective rather than how a particular article affects its readers. This work focuses on a reader’s stance, one forms after reading a certain content. Ontology driven knowledge base is used for semantic matching to further calculate the coherence of the words to distinct intermediary emotions. On reception of a new document, frequency of terms are calculated which are then matched with ontologies and hence classification is done using deep learning-based classifier. A series of experiments are taken up on these news documents, and hence an inference for the proposed method is marked out for it being much reliable than other existing systems for emotion detection of a text document.