About a month ago I blogged about the question “ Do you need a Controlled Vocabulary?”
The post was a response to a statement from a client who doubted that people in his section of his company needed a controlled vocabulary, because he was sure that everyone used terms and names consistently.
I told him that it seemed unlikely, and went on in the blog entry with an example of a well known organisation in London (The National Gallery on Trafalgar Sq) that can’t seem to keep their own name consistent in their podcasts.
He promised me that he’d get some data and get back to me.
Well, he did, and confirmed what I expected – they could really use a controlled vocabulary.
His company produces a number of software products. During development, each product has a code name (drawn from a hat, I’m sure, because it bears no apparent relationship to the final name) and a release name. As it happens, the final release name is often different from the initial release name and the spelling of the final release is often unconventional, so that it can be trademarked.
Not surprisingly, people find this difficult to keep straight. They’ve been starting to tag information with metadata to help them find their information for internal reuse. Unfortunately, the tagging has been inconsistent, with some people using the right name, others using old names and as often as not, people use the traditional spelling of the name rather than the trademarked spelling. A few didn’t even bother to provide proper tags, and just entered “asdf” as a tag wherever they needed a tag. Personally, I don’t think that these few were taking the exercise seriously at all…
When he looked a bit deeper, he also found that because some software used modules from other software packages, people tended to mix and match names when they were describing the products.
In a case such as this a controlled vocabulary helps. It ensures that only correctly spelled product names are entered, it can link pre-release code names with release name, and it keeps people from entering random characters as tags. It doesn’t solve all the problems with tagging, but it does make it more likely that the tags will be useful.