Inter country poetry classification using Topic modeling

Poetry is an art of arranging the carefully picked words in a specific order to express the authors experience and emotions. Poetry in India has its strong roots with world famous and excellent poets, amongst few renowned poets are “Universal Poet” Rabindranath Tagore, “Nightingale of India” Sarojini Naidu and “Swami” Vivekananda. Poetry style varies from country to country depending on the author. Author’s poetry topics, words and style depends on the circumstances they raised in, situations they faced, and their mind set. Many authors who belongs to India but settled in western countries and written their poems, in this context automatically identifying a poem’s author is a challenging task for the literary scholars who analyze the poetry. In this work authors proposed a method based on Latent Dirichlet Allocation(LDA) topic modeling to classify the poetry written by Indian or western(American)poet based on the distribution of topics per document. The experiment is performed on 3 data sets 128, 1600 and author wise poems respectively. This experiment is performed based on semantic features. Best result 91% precision and 88% accuracy is achieved on author wise poems data set using random forest algorithm