For the past eight years, we have been focusing on taxonomy based definitions of our content and our site map. The internal taxonomy that we have developed is based on industry use of specific terms, as well as coined terms that define IBM products and services. With My developerWorks, we will be introducing a social bookmarking capability, as well as tagging functionality for our own content, that will allow users to tag and self-identify developerWorks assets. The end result is that we will have a hybrid approach that will make use of both capabilities.
I say assets here and not content, because one important capability with My developerWorks will be the ability tag yourself when you create a profile on developerWorks. In addition, there are taxonomy listings that allow you to identify your skills, roles and interests.
Why the combination you ask?
Each approach (taxonomy and folksonomy) has its strengths and weaknesses. Taxonomies are valuable because its easier to understand a body of content if you have a common understanding of what terms mean. You can establish structure and get reliable results based on your searches, given that there is a common understanding between identifiers and searchers.
The breakdown usually comes in just that. There is a difference in interpretation, and as a result, what the searcher looks for is not what the identifier has tagged things as. The end result is that the searcher surrenders and moves to a full text approach, thereby nullifying work efforts that go into tagging.
On the other hand there is folksonomy. With this approach, each user can tag or bookmark things exactly as they wish. What gets lost is the standardized approach, and reliability of a set of structured and previously agreed upon values.
The availability of both aspects in the My developerWorks profile will allow us to provide maximum functionality when it comes to one of the key features we heard requested for social tooling - the ability to find and connect to peers and experts. When you use tags, you can put in any value. So when I see a list like the following, it tells me some things, but not others.
I know from this list that the person (lets call him Ravi) wants to be found for all these things. However, I do not know what Ravi's intention is for these things.
Consider instead, Ravi entering these interests using our taxonomy lists, where we ask about interests, skills, and role. The result might be the following info:
The screen would look something like this:
with this list, we know what Ravi knows, and what he is interested in (which probably means he wants to learn it). Given that, we can do a lot more when it comes to providing recommendations. We will add Ravi to lists of people looking for Java developers, and show him lists of people who list 'python' in their skills.
So in summary, the intention is to continue to maintain the taxonomy work, because that is a valuable contribution to the Information Architecture realm. At the same time, we have a valuable braintrust in the combined experience of the 3.1 million unique visitors who can add a lot of value to developerWorks content by tagging it and making it more useful, relevant and findable by their peers.
Tagging; at the heart of people search for My developerWorks
brenny 110000SGFR Tags:  connections mydeveloperworks web2.0 tagging taxonomy folksonomy 1,262 Visits