Sunday, September 9, 2007

Concept Maps or Topic Maps

Need Some Help...

So in furthering this idea of enabling a collaborative space for users to visually map patterns to construct relevant pattern languages, I have been considering the various approaches afforded through Concept Maps and Mind Maps. Now, each mapping approach has a list of associated software products to aid in this visual process. Some of these products such as Cmap provide a server tool to allow for collaborative mapping among multiple people, basically providing exactly what I have been looking to implement for the patterns we have.

This seems like a very possible avenue to take. Plus, it is free to educational institutions and they provide donations to non-profits, like PSP. For those unfamiliar with idea of concept maps, see this definition:

"Concept mapping is a technique for visualizing the relationships among different concepts. A concept map is a diagram showing the relationships among concepts. Concepts are connected with labeled arrows, in a downward-branching hierarchical structure. The relationship between concepts is articulated in linking phrases, e.g., "gives rise to", "results in", "is required by," or "contributes to" - Courtesy of Wikipedia.

Basically, these concepts are linked together in a semantic relationship defined by the person or persons designing the map. So, for us pattern language designers we could in focus on encouraging the development of a single pattern and define semantic connections to other patterns below it and above it in the pattern language heirachy.

Take a look at the image below and envision, say "Grassroots Public Policy Development" as taking the place of what is currently there. Similar to the map below we could make connections and assert that Grassroots Public Policy Development is enabled through greater levels of "Civic Capabilities," which is dependent upon a certain level of "Civic Intelligence." Civic Intelligence might also be supported through patterns such as "Informal Learning Groups" and continued access to "Political Settings."

However, some might say that this approach is limiting and some ways it might be. I have been reading a bit on Topic Maps and came across this response on infoloom to the question surrounding the difference between concept maps and topic maps:

"First, concept maps and topic maps are born out of the same desire: to
represent what we think or what we know in a graph. Here, the term graph
can be taken both in its visual sense, and in its mathematical
sense. Concept maps emphasize the visual and, in the paper and pencil
instance, ignore the mathematical sense. Topic maps emphasize the graph
structure, the mathematical sense, and leave the visual or presentation
entirely up to implementors of topic map engines.

The second response is to delve into the structures of concept and topic
maps. They are both the same and they are different. Concept maps, as I
think Joseph Novak formalized them, consist of named nodes and labeled
arcs. Topic maps, as I think Steve Newcomb and others formalized them,
consist of a few more objects than that, objects that deal with
occurrences, scopes, roles, association membership, and so forth.
- Jack Park

I have been reading other responses in other areas and much of what is see appears to be similar. Now, there seems to be something interesting here in that it is possible that we don't want to build concept maps at all but rather topic maps, that could possess the visual structure of a concept map. Perhaps this is what we are after in designing these pattern languages through the web. Perhaps, the extension of these maps to include the scope, associations and so forth.

I would really like to here some thoughts on this...


Yishay Mor said...

I'm not familiar with topic maps, but we used FreeMind mind maps extensively in the Learning Patterns project.

Concept maps seem to be semantically richer, MindMaps are inherently hierarchically. The richness of concept maps might come at the expense of readability / search / navigation. MindMaps lack the facility for structured lateral links. Ideally, I would like to see an extension of MindMaps which allows me to express specific types of cross-links, such as 'follows', 'uses' etc.

Justin-G said...

Well, that is what I was trying to ask the other day when chatted, whether or not MindMaps allow for semantic linking.

As it stands I think MindMaps are pretty, but to me they are less readable in the sense that I'm unsure as to how pieces interact or how they are connected.

Looking at Novak's work and the types of examples he has generated regarding concept maps it seems to me that there is a basic order that one can follow as if reading a list of sentences about some interconnected concepts.

This is one reason that I used a concept map approach versus mind-maps in my thesis; mind-maps are pretty but (for me) takes a lot for me to get the meaning.

Now, the funny thing is that most people I talk to seem to find the opposite and perhaps that is highlighting some learning differences surrounding visual versus linguistic centered. I like both, but for some the injection of semantic relationships might blur the overall structure.

When looking at patterns I have to be able to conceptualize how they link. So simple words as you say, 'follows,' 'uses' or 'enables,' 'supports' brings a whole new dimension necessary for understanding systems based design. Plus, it is easier to then critique these assumptions and make adjustments based upon empirical observations.

Whereas without that additional information it is difficult to know what one intended after the fact when they asserted some sort of relationship.