Low Dimensional Visualization of Folk Music Systems Using the Self Organizing Cloud

We describe a computational method derived from self organizing mapping and multidimensional scaling algorithms for automatic classification and visual clustering of large vector databases. Testing the method on a large corpus of folksongs we have found that the performance of the classification and topological clustering was significantly improved compared to current techniques. Applying the method to an analysis of the connections of 31 Eurasian and North-American folk music cultures, a clearly interpretable system of musical connections was revealed. The results show the relevance of the musical language groups in the oral tradition of the humanity.