Category analysis

As a participant in the test categorizations, I'd be really interested to see details of that data. Which categories were the most distinct? Were there certain types of books with a lot of overlap? Did certain types of books shift categories? I think it would be helpful to see the data to understand the kinds of issues to consider for the next round. After all, talk threads are one thing, but categorization is another. For example, I noticed that although BISAC generally puts meditation in the mind/body/spirit category, OSC proposes to remove it from the renamed "paranormal" and put it into health. Should Wicca books classified with religion or with paranormal; what did the data say? I'm pretty sure science doesn't accept claims of any religion, no matter how popular, but we don't put all that in paranormal. My bookstore groups Medicine with health rather than science; did the data test point the other way? I did read the Thingology post, but I think given the amount of time and effort put into testing more sharing of details would be helpful. Not just the results, but the decision-making data and process. Openness is key in any supposedly collaborative effort, as OCLC is discovering as of late.

Good questions all!

Another way in which the first round classification data would be useful is with some tag analysis. For instance, if we could have the books that were classified as "religion" and then find out the tags used on them, with popularity of tags, and tag combinations, then that might help us think about the next level of categorisation for the "religion" category.

Hello. We are waiting to get a break down of the testing data from Tim. As soon as we get it we will post it to the wiki.

I think what would be very helpful is to create a sort of mock library of 1,000 or so non-fiction volumes that represent the typical holdings of a public library. From that we can arrange and re-arrange and see how these categories and sub-categories actually divide the books.

When we all work with our own imaginary libraries, we end up with entirely different conclusions shaded by thoughts of our own home collections and our own biases.

>5 mattsya:

A 1000 is way too small. The smallest libraries have over 10,000 books. I, however, agree with the problem. I think a lot of people are in this game to classify their own books, not to provide public library access.

One solution might be to use the collection at one of the LibraryThing public libraries and their online catalog as a base collection.

Or everyone could go to their local public library and browse the stacks. Try to see what works and what doesn't.