Classification of Indian Classical Music Ragas using Graph Theory

A Raga is a melodic framework of a certain structure of notes. The Ragas of Indian classical music are roughly divided into 10 divisions by V.N. Bhatkhande based on certain rules and specific melodic styles. This is known as the traditional ’ Thaat’ system. This study aims at grouping Ragas using graph clustering methods and compare the results with the traditional methods of classification. This classification is based on the similarity between ragas on basis of 35 parameters based on the presence and absence of notes and characteristic notes. In this study, graphs are classified with two methods. One of them is a community detection algorithm using mixture models. The other method is based on classification using the nearest neighbour where traditional group members are selected as the classification criterion. In this study, the aim is not classifying music into ragas, on which a lot of research has been previously done, but classifying ragas themselves into groups. Our results match well with the traditional methods of classification.