Is There a Problem Here?

In looking over what we've got so far, I can't thinking that this system isn't quite doing what it should be. Would the entry on bluegrass be helpful to someone who was unfamiliar with the style? Further, is that the idea of this project to begin with?

At first, I thought that the values in the fields are too concise; there's not enough detail to explain how all of the characteristics that make up bluegrass fit together. So would it be better to have fewer fields but have longer, prose, explanations in each? My fear there is that records will begin to look like Wikipedia entries. There would be no point is duplicating what's already been done by Wikipedia; that certainly is not the purpose here.

So what is? The concerns I'm having seem to bring up fundamental questions about Universal Music Classification. What exactly is the point of the system? How do I best set it up to accomplish its purpose, once I'm clear on what that is?

What I had initially envisioned is a bibliographic tool; a system that would help librarians or music collectors organize music by style. It is intended to do that by telling you what style a given piece of music is by looking at its musical characteristics. Accordingly, a prose explanation of the style is not what is needed; again, Wikipedia has that covered already. Instead, I'm seeing two possible forms that this could take that would make it a useful tools for classification.

The first would be as a sort of flow chart; a series of questions about the given piece that would ultimately lead the user to the proper genre. It might start by asking whether there are any electric instruments; if so, then the system would rule out acoustic-only styles like bluegrass or anything pre-twentieth-century. Questions could be designed that would narrow the possibilities until only one possible styles remains; the piece would then have to be that style.

There is, however, an issue with this. One of the problems that catalogers face in terms of recorded music in particular is that the information object (tape, CD, etc) and its packaging often have limited information about the music itself. What if the next question the system presented you with was about the music's time frame of origin, but you didn't know that information about it? A flow chart has the potential for dead ends.

The second possibility would be set up a list of words/concepts associated with each style. So your list of words to go with Rock would include things like: guitar, drums, electric, loud, backbeat, etc. In other words, there would be a list of ideas associated with Rock. This way, the user could enter a keyword based on what was known about the music, and the system would return all of the possible styles that are associated with that keyword. The user could continue entering known characteristics of the music until the system returns only one style that has all of the given characteristics.

What each of the above ideas hinges on is not what makes up a style, as much as what distinguishes one style from another; that's what this system is really all about. I'm going to look at each of these ideas more closely and try to decide which approach would work better (or whether some combination of the two might be possible). This will clearly alter the system in a pretty fundamental way, so it's not a decision to take lightly. As always, I appreciate any feedback you might have.