[r-t] Composition database

[Graham John]

I'm back from hols and hope you enjoyed the respite from my messages. I had
mistakenly hoped that you would have completed the new definition of a peal,
or at least added something to the Changeringing Wiki by my return!

Don said (in thread Leary's 23-spliced):

> For things being newly submitted the format is not the major
> stumbling block. Computer proof, and in a form that emits the
> form that will be uploaded, is the major stumbling block
> there. It's startling the number of things folks submit that
> have errors of one sort or another.

> The long term solution to this is to make proving software
> available that generates things in the right format.

A composition database and software for input and interrogation has been
long been needed. As a guest, I presented a paper on the subject to one of
the CC committees (probably PCC but might have been MC or CCC) in the early
1990s.

At that time I listed the possible Objectives of a Composition Database and
associated software as:-

For the Peal Compositions Committee:

*	Provide a master reference collection of all compositions
*	Confirm uniqueness and establish authorship
*	Register every composition, assigning a unique reference
*	Confirm truth
*	Set for publication directly from the database
*	Easily respond to requests for compositions
*	Prepare collections for paper publication
*	Provide a mechanism for inputting and maintaining collections
*	Compare and contrast qualities of different compositions for reviews

*	Identify duplicates, rotations and variations 

For Conductors:

*	Find suitable compositions for a method given particular
requirements
*	Identify compositions suitable for a method even if not specifically
composed for it 
*	Compare and contrast qualities of different compositions	

For Composers:

*	Check uniqueness of a new composition
*	Submit compositions
*	Provide a comprehensive single reference work for compositions
*	Compare relative merits of existing and new compositions
*	Identify gaps in the collection 
*	Identify which methods a new composition works well for

For Historians:

*	Provide cross references for sources of publication and reviews of
compositions

The XML format and ringing.org composition collection is good as far as it
goes, but the main disadvantage is that the "calling" is still held in an
undefined display format, even though it may have been generated by a
proving program in the first place. This means that the current collection
of compositions cannot be queried intelligently.

For example, new methods are being rung all the time. I am sure that
conductors would love to know which compositions in the collection will work
for it, and if they do, what music would be generated. What is needed is the
ability to select a method from the method library, or enter a new method
place notation, and run a query that proves all suitable compositions
against that method with selection / music analysis & scoring to the
conductor's taste. Provision of software like this could reduce the ringing
of inferior BYROC and SMC32 generated compositions, when better compositions
are already available (and much quicker to find than by exhaustive search). 

Another example is when a composer wants to know if a composition they have
produced has already been published by someone else, perhaps for a different
method. They would input their calling, and run a query which could use
various algorithms (calling position counts, reversals, rotations, QSet
comparisons etc) to identify potential trivial variations for the composer
to inspect. If they are happy that it is new, they can submit it to the
database, pre-validated, and instantly available for others.

While on holiday, I gave some further thought on how this could be done, and
will report back when I have produced a prototype. 

Graham