Two text cloud representations of the 43 knowledge management definitions.
Before processing I removed the phrase ‘knowledge management’ when used as “knowledge management is…” and I merged British and American spelling variations for organization (’s’ versus ‘z’).
I’m still digesting and also thinking about the other analysis ideas mentioned in the original post. What do you make of these?
From TagCrowd:
access activities application approach assets best business collective company creating creation data documents effective efforts employees enterprise experience explicit gathers human identifying improve individual information innovative intellectual km knowledge leveraging making management objectives organization organizational people practices process retrieval sharing strategy system systematic tacit technology terms transfer used value workers
created at TagCrowd.com
From IBM’s Many Eyes, an interactive tag cloud:
Well, being a very visual person, I find both of the interesting but I’m not sure what they tell me in this context. The presentations are not a “definition” per se, but they highlight what seem to be important parts of the many definitions.
There are almost too many words to focus – it looks like most of what I’d think of as “stop words” in search terminology have been removed (“and”, “of”, “the”, etc), but perhaps raise the threshold for inclusion a bit to see what you get (obviously, the more commonly-used words but does it change the immediate impression)?
Perhaps you could try a “social network” where words are connected in the network if they are, say, within 2 or 3 words of each other in any definition and the strength of the link is the number of definitions in which the words are close and the size of a node is the number of times a word is used? That might provide more insight about what words are used often and also commonly used in conjunction.
The tag cloud as provided presents little more insight than a sorted list of the words (ordered by their use).
Just my two cents.
I was trying to build a sentence based on the weight of the tags, when I quickly realized that the weight of the tags won’t provide an order for meaning. The two-word cloud is a much more even spread making it more difficult to find any insight.
I like the clouds a lot. They put K in the middle of KM surrounded by all the words that you would expect to see but without a sentence to add false and distracting structure.
Great research effort, Ray. Keep it coming! This is valuable and interesting to all of us. Maybe two different tag clouds would be helpful…one on the “people/process” part of KM and one on the “info/content” part. Thanks for making this readily available to people.
Lee, Shaun, Matthew, and Kaye: thanks for giving a go at working with these and for the feedback.
I think Matthew hit it on the extent of the value…the text clouds show “aboutness” (including “more or less about”) without bogging back down into sentence structure. For this I prefer the TagCrowd version; although, that is not really an apples-to-apples comparison since in TagCrowd I experimented a bit with the number of words and settled on 65 as what gave the most digestible story for me without bringing more distraction from the core concepts. I did not explore if this was a parameter I could change in ManyEyes.
If anyone wants to work with this further, the dataset is available via ManyEyes (for the 43, not the now 53, definitions) or I could email you the text file.
Cory Banks also used ManyEyes on the definitions. Not mentioned in his blog post is this word tree that he also created.
Matthias Melcher created a topic map from the definitions using the DeepaMehta tool.