We propose an algorithm for exploiting statistical
properties of large-scale metadata databases about music
titles to answer musicological queries. We introduce two
inference schemes called “direct” and “inverse” inference,
based on an efficient implementation of a kernel
regression approach. We describe an evaluation
experiment conducted on a large-scale database of fine-
grained musical metadata. We use this database to train
the direct inference algorithm, test it, and also to identify
the optimal parameters of the algorithm. The inverse
inference algorithm is based on the direct inference
algorithm. We illustrate it with some examples.
