Bias in Classification

Sitting in class right now, I'm contemplating the issues that surround the inherent biases in classification systems. I think there are strong arguments that any specialized group of people would benefit from a custom-fitted classifiaction system for organizing and retreiving information; however, I don't think it's possible to create a universally useful system.

In order to create such a system without any bias, the system would have to be composed of an entirely arbitrary classification system (...and who is that useful to? No one!).

I'm wondering if a database sorting system could help with this. For example, every work in a database is assigned an ID number. It doesn't matter how te numbers are assigned. Then, librarians could go through and customize organization systems for specialized groups of users by using a tagging system to identify works of a group and assigning labels to the categories. This might work similar to the way that Pandora has music specialists assigning works in relation to each other.

Then, users could come in and pick the organizational schema that fits their needs best, or possibly even create their own.