[r-t] Distances in composition space

[Graham John]

MDB wrote:

> Anyone got any comments on this, or got any better ideas?

Score against multiple criteria, such as:-
 
1. Composition Length

Need to think here are about how to cover variations that are simple
reductions or extensions.

2. Count the calls, and types of calls by calling position

Need to check for reversals and rotations. Duplicates are often created
unknowingly for different methods.

3. Check the proportion of rows that are common

Need to check for reversals and rotations.

4. Check whether the commonly accepted musical properties are similar

Although it is possible to have compositions with similar mucial properties
with significantly different callings.

The sensitivity of the scoring could be adjusted to identify a small number
of possible matches that could be inspected visually. Although the treatment
of tvs can be rather subjective, there are probably a significant number of
straight duplicates published, so the ability to automatically process a
database of compositions against a new one would be valuable.

Graham