Literary Style Determination Dased on Statistical Hypothesis Testing and KNN Approach

Abstract

The paper presents a method for the literary style determination. It is based on a re-sampling approach and character level features. A text is considered as a sequence of characters (n-grams) generated by different random sources. Bootstap-like approach is used to draw samples from the texts. Kolmogorov-Smirnov two-sample test and KNN based statistic are applied. Experiments with texts in English and Russian are given, illustrating the algorithm operation.