As information architecture tends to be an emerging field (though gaining more and more recognition among many mainstream companies and organizations as a core functional area), I like to have friends and family make a guess at what IA is all about before telling them. My brother, who happens to be a traditional architect (he designs buildings) said 'hmm...' then gave a definition of IA that I wish I had captured on tape to use as an elevator pitch. The real meaning of the "architecture" part gets lost on most of us IT types and occasionally we need to pause and rethink IA in order to avoid the scope creep into the world of Web development, UI design and everything else that has no direct connection to the information itself (or just addresses it on the surface).
For different information artifacts (needles) to be associated with each other throughout an information system (the haystack), they need to share common identifiers. I know the DBAs reading this are thinking 'duh,' that's why data in databases have things like keys, foreign keys and other identifiers that help us locate objects and associate them with each other. Well, guess what? The same logic holds for Web content and documents in an enterprise. Common identifiers are the strings that tie information needles together in the haystack, making them easier to aggregate and retrieve. (Hence, the name of this site...doesn't seem as off-topic now, right?)
What exactly are these common identifiers? They are nothing but tags (numbers, letters, words..). Information needles are virtually strung together when they use the same tags. There are two ways of achieving this. You can set up a search engine, flip on the switch and hope that 'like' information happens to use the same words, and more importantly, the words that most people will use to look for them. Alternatively, you can pre-define a core set of tags to use to ID different types of information then apply those tags to them as common identifiers. Using the second method, you can string the needles together however you need them aggregated and be guaranteed to find them whenever you want them.
Now, the non-DBAs reading this might be thinking "well that sounds pretty simple. Where's the rub?" The rub is that, despite how straightforward this sounds, enterprises struggle to come up with a satisfactory set of tags (controlled vocabularies, taxonomies, ontologies, etc) to use as the common identifiers for their information. The ones who achieve this first step face many challenges trying to apply the tags to their information artifacts in a consistent and seamless manner. When and how should organizations automate the application of tags? Should all or some of this tagging be done manually? How do organizations get employees to consistently apply the tags that cannot be automated? Visit again (or subscribe to this blog via RSS) for some answers to these questions, or comment on this blog if you already know the answers!