Cognitive Adequacy in the Measurement of Melodic Similarity: Algorithmic vs. Human Judgments