A concept map is a knowledge model, represented as a labeled set of nodes and arcs used to summarize a body of knowledge on a topic, much like an outline. Each node typically represents a concept, and an arc between two nodes represents a relationship between two concepts. In this paper we will introduce the concept map model and then survey and compare three concept mapping tools.
A concept map is comprised of concepts and relationships. There are no restrictions as to the range, scope or topical focus of a concept map.
|Figure 1.||Illustration of a directed relationship between two concepts.|
In general, the concept map model places no inherent restrictions on what can or cannot be a "concept". A concept is much like the Platonic "idea" or the "subject" in predicate logic: almost anything can be considered a concept. Some concept mapping tools allow users to define different types of concepts to further restrict the model, but these restrictions are imposed internally.
Relationships indicate a conceptual connection between two nodes or concepts in a concept map. A relationship can be directed or undirected. A directed relationship points from one node to another. As illustrated in Figure 1, two nodes labeled Apples and Trees might have an arc between them called "grow on" with the arc directed from Apples to Trees, indicating that Apples grow on Trees.
In this section, we compare three concept mapping tools, GetSmart, SMART Ideas and CmapTools. We will first introduce each, and then compare them in the context of various features and characteristics.1] Lead by Dr. Hsinchun Chen, GetSmart is a collaboration between the AI Lab at University of Arizona and Virginia Tech.
|Figure 2.||GetSmart screenshot, a flowers concept map.|
GetSmart comes in two versions, stand-alone and server dependent.
The stand-alone version is a simple concept mapper, capable of drawing lines and boxes, and not much else. However, there is an additional module that enables searching the local repository for keywords.
Since GetSmart is programmed in Java, it can run on any operating system that supports the Java virtual machine.
GetSmart has a built-in taxonomy of relationships. While the user is free to label relationships in whatever manner they see fit, the GetSmart taxonomy provides a few generally applicable relationship names to choose from (See Figure 3).
|Figure 3.||GetSmart's taxonomy of relationsihps.|
Each concept can be associated with zero or more URLs or files (Figure 4). Thus each concept can be further described by these associated resources. This ability to associate URLs or files with a concept can be viewed as a simple form of superimposed information where the files or web pages are described by or linked with certain concepts. Superimposed information is an approach to information representation and information management under investigation led by Dr. David Maier and Lois Delcambre where information from a base layer can be easily referenced from the superimposed layer. In this context, the relationships in a concept map could be used to describe how base-layer resources are related.
|Figure 4.||Associating a concept with a resource.|
The server-dependent version of Getsmart is a web-based application, that works the same as the stand-alone version, but is deployed as a Java Applet. Clients interact with a shared concept map server via a web browser, allowing users to store concept maps in a central repository of concept maps and look at each other's concept maps. The web-based version is deployed as a web-based service, and access to the tool can be requested by visiting http://boris.dlib.vt.edu:8080/aicm/index.html (Figure 5).
|Figure 5.||The GetSmart web interface, for collaborative concept mapping.|
The most feature-rich GUI among the three concept mapping tools is SMART Ideas. It has an elaborate selection of options for controlling the appearance of concept maps. The style pallet provides a number of options for the appearance of concepts and relationships. It offers a small but extendable library of clipart, so that in addition to the usual circles and boxes, concepts can be represented as images.
SMART Ideas also provides a number of extensions to the core concept-map model. Multi-level diagrams allow concepts to have relationships with other concept maps. Concepts can also link to files, URLs and attachments. Cliplets are interactive widgets similar to Java Applets. For example, a cliplet could be a fully functional timer, a pair of dice, an actual working clocks, or a functional calculator. A minimal API is available for developers to write their own cliplets.
The template library offers a nice showcase of what can be done with SMART Ideas, and can serve as a good spring-board for creating a visually appealing concept map.
Finally, SMART Ideas saves concept maps in their own proprietary format, but can export to Microsoft Word format.
|Figure 6.||SMART Ideas screenshot, illustrating the order in which clothing must be put on.
A concept represents a piece of clothing, a relationship represents
a dependancy on another piece of clothing.
Cmap Tools offer nice looking and feature rich GUI. Concepts can be visually styled with colors, shadows, background images, etc (Figure 7). The auto-format feature can automatically position concepts relative to each other, to make a well-proportioned map. Concept maps can be exported in a wide variety of formats: as an image, a web page, an outline, propositions as text, an XML document or a LifeMap.
Cmap Tools is programmed in Java and run on Windows, OSX, Linux and Solaris.
|Figure 7.||Cmap Tools screenshot, a design decision concept map, retrieved from the
collaborative concept map space, this one from a class at University of Massachusets.
Like GetSmart, Cmap Tools can connect to a concept map repository for collaborative management. Cmap Tools is being used by a number of universities as a repository for classroom concept maps. The collaboration features are quite advanced, allowing the user to search across the entire repository of concept maps, as well as join a "Soup", a shared repository of concept maps that can be retrieved. The system allows a user to sign up for an account, share concept maps, and participate in a integrated threaded-messaging forum for the discussion of a particular concept map (Figure 8).
|Figure 8.||Cmap Tools repository, authenticating against the repository for setting up a shared space.|
In this section, we identify various axes for comparison among these concept mapping tools, describe each axis and then analyze each tool.
First we list the tools, the license under which the software is released, the organization responsible for their creation and maintenance, the operating system environments supported by the tool, and what formats maps can be exported as.
|GetSmart||Open Source?||AI lab at the University of Arizona, Virginia Tech||Java (any OS)||XML|
|CMAP Tools||Commercial||Institute for Human and Machine Cognition||Windows, OSX, Linux, Solaris (Sparc)||Image, Web Page, Cmap Outline, XML, Life Map|
|SMART Ideas||Commercial||SMART Technologies||Windows, OSX||Microsoft Word|
Shapes and colors can be good indicators of additional information about concepts. A good coloring scheme can convey additional dimensions of information beyond simple concepts and relationships. For example, in a concpet map of fruits and vegetables, one might color all fruits red and all vegetabls green, or use circles for fruits and squares for vegetables. In this section we address the visual styling elements that can be added to a concept node.
|Software||Colors||Shapes||Background Images||Shadows||Borders||Text Align|
|GetSmart||supported||squares||not supported||not supported||not supported||center|
|CMAP Tools||supported||squares, circles||supported||supported||supported||center, right, left|
|SMART Ideas||supported||squares, circles, triangles, rounded boxes, etc.||supported||supported||supported||center, right, left, top, bottom|
Similar to concepts, color can indicate additional information about relationships. This section compares what types of relationships are supported, and how they can be enhanced visually.
|GetSmart||supported||supported||not supported||supported||(blank)||URLs, Files (broken?)|
|CMAP Tools||supported||supported||supported||supported||(blank)||URLs, Files|
|SMART Ideas||supported||supported||supported||supported||supported||URLs, Files, Attachments|
This section compares features that apply to an entire map. Colors says whether or not the background color can be changed. Background patterns can be added within some tools as well. Multilevel maps are can link from a concept to anothre map. Autolayout features can take a cluster of concepts and lay them out in a balanced graph. Quick-connect mode allows the fast creation of many conecpts with a single-click per concept.
|GetSmart||supported||not supported||not supported||not supported||not supported|
|CMAP Tools||supported||not supported||not supported||supported||not supported|
|CMAP Tools||supported||free account||supported||supported||supported|
|SMART Ideas||not supported||n/a||n/a||n/a||n/a|
In this paper we briefly surveyed three concept mapping tools, and compared their features. GetSmart proved to be a simple, clean, cross-platform concept mapping tool. IHMC's CMapTools is a good concept mapping tool for those interested in sharing maps across repositories and organizations, and delivers a number of visual styling options. SmartIDEAS offers a wealth of features for creating visually appealing concept maps, and multi-level concept maps.