Poetry chronological classification: Hafez

Abstract

We propose a novel task, the chronological classification of poems. Houman’s research methodology, which we have adopted here, has provided us with labels for the ghazals that we used as training data in order to train automatic classifiers to classify the rest of the ghazals. To achieve our objective, we have prepared a corpus of Hafez poems in digital form with consistent idiosyncratic linguistic behaviors. Our classification framework uses a combination of unsupervised and supervised methods: a Support Vector Machine (SVM) classifier that uses Latent Dirichlet Allocation (LDA)-based similarity features.