Bilingual Chronological Classification of Hafez's Poems

摘要

We present a novel task: the chronological classification of Hafez's poems (ghazals). We compiled a bilingual corpus in digital form, with consistent idiosyncratic properties. We have used Hooman's labeled ghazals in order to train automatic classifiers to classify the remaining ghazals. Our classification framework uses a Support Vector Machine (SVM) classifier with similarity features based on Latent Dirichlet Allocation (LDA). In our analysis of the results we use the LDA topics' main terms that are passed on to a Principal Component Analysis (PCA) module.