Computational analysis of dystopian elements in the partition fiction: A machine learning approach to the indian English novels

Abstract

Despite the growth of digital humanities, the specific application of computational methods to analyse literary themes and elements remains underexplored. This study aims to use machine learning algorithms for text classification on seven novels about the partition of India. The article has a dual objective: firstly, to develop and implement an innovative opinion mining framework that leverages machine learning techniques to identify and classify dystopian elements in Partition novels; and secondly, to evaluate and compare the efficacy of the model in accurately classifying dystopian sentiments within selected literary texts. The proposed framework includes six phases of data collection, test-pre-processing, text extraction, text exploration, modelling or classification using machine leaning, and performance evaluation. Machine learning approaches such as Logistic Regression, Naïve Bayes, Support Vector Machine, and k-Nearest Neighbor are adopted to classify the text based on compiled dictionary. First, the study revealed that the main components and characteristics of dystopian elements found in selected narratives are ‘fear’, ‘suffer’, ‘oppression’ and ‘violence’. Second, the analysis showed that SVM with CountVectorizer, using imbalanced dataset and random oversampling dataset, outperforms other classifiers in classifying dystopian types in the selected novels. The results also suggest that CountVectorizer works better for the dataset compared to TF-IDF.