24.7 Semantic Networking as cognitive tools

24.7.1 What Are Semantic Networks?

A new genre of cognitive tools, semantic networking tools, has appeared in recent years. Programs such as Sem.Net (Fisher, 1990, 1992), Learning Tool (Kozma, 1987, 1992), and TextVision (Kommers, 1989) are cognitive tools that provide visual and verbal screen tools for developing concept maps, otherwise known as cognitive maps (see Fig. 24-1). Cognitive maps are spatial representations of ideas and their interrelationships that are stored in memory. These tools enable learners to interrelate the ideas they are studying as multidimensional networks of concepts, to label the relationships between those concepts, and to describe the nature of the relationships between all of the ideas in the network.

Semantic networks are representations of human memory structures. The cognitive theory underlying semantic networks maintains that human memory is organized semantically, that is, according to meaningful relationships between ideas in memory. These ideas, known as schemas are arranged in networks of interrelated ideas known as semantic networks. Semantic networking programs are computer-based, visualizing tools for representing semantic networks. Perhaps the best-known theory of semantic networks is active structural networks Quillian, 1968). These are mental structures composed of nodes (representing schemas) and ordered relationships or links connecting them. The nodes are instances of concepts or propositions, and the links describe the prepositional relationship between the nodes. In computer-based semantic networks, nodes are represented as information blocks or cards and the links as labeled lines (see Fig. 24-1).

The purpose of semantic networks is to represent the organization of ideas that someone knows about some phenomenon (e.g., baseball) or the underlying organization of ideas in a content domain (e.g., sociology). Semantic networks function as cognitive tools by engaging learners in analyzing the structural relationships among the content being studied. They can also be used as evaluation tools for assessing changes in thinking by learners (Preece, 1976). If we agree that a meaningful representation of memory is a semantic network, then learning can be thought of as a reorganization of semantic memory. Producing semantic networks reflect those changes in semantic memory, since the networks describe what a learner knows. In this way, semantic networking programs can be used to reflect knowledge acquisition.

24.7.2 How Are Semantic Networks Used as Cognitive Tools?

Semantic networking aids learning by requiring learners to analyze the underlying structure of ideas they are studying. The process of creating semantic networks engages learners in an analysis of their own knowledge structures, which helps them integrate new knowledge into existing knowledge structures. The result is that the knowledge that is acquired can be used more effectively. Kozma (1987, 1992), one of the developers of the semantic networking tool Learning Tool, believes that semantic networks are cognitive tools that amplify, extend, and enhance human cognition. Constructing computer-based semantic nets engages learners in (1) the reorganization of knowledge through the explicit description of concepts and their interrelationships; (2) deep processing of knowledge, which promotes better remembering, retrieval, and the ability to apply knowledge in new situations; (3) relating new concepts to existing concepts and ideas which improves understanding (Davis, 1990); and (4) spatial learning through the spatial representation of concepts within an area of study (Fisher, Faletti, Patterson, Lipson, Thornton & Spring, 1990).

24.7.3 What Research Supports the Use of Semantic Networks as Cognitive Tools?

The usefulness of semantic nets and concepts maps is perhaps best indicated by their relationships to other forms of higher-order thinking. They have been significantly related to formal reasoning in chemistry (Schreiber & Abegg, 1991) and reasoning ability in biology (Briscoe & LeMaster, 1991; Mikulecky, 1988). Semantic networks can also provide a useful evaluation tool for measuring the acquisition of knowledge. In a geometry class, concept maps were used to evaluate teaching outcomes and to monitor student progress in the course (Mansfield & Happs, 199 1).

An important research agenda in learning psychology focuses on the expert-novice distinction, comparing student knowledge representation with teacher or expert representations. Research has shown that during the process of learning, the learner's knowledge structure begins to resemble the knowledge structures of the instructors, and the degree of similarity is a good predictor of classroom examination performance (Diekhoff, 1983; Shavelson, 1972, 1974; Thro, 1978). Instruction, then,. may be conceived of as the mapping of subject-matter knowledge (usually that possessed by the teacher or expert) onto the learner's knowledge structure. Semantic nets are a way of measuring that convergence.

Using compared Pathfinder nets (Schvaneveldt, 1990), researchers have also shown that semantic nets are related to course examination performance (Goldsmith, Johnson & Acton, 1991). In a study examining the use of generating computer-based semantic networks in a computer programming course, Feghali (1991) found that students who built nets scored better in course tests; however, the differences were not statistically significant. More and better research of this type is needed to verify a consistent relationship between particular criteria for evaluating semantic nets and traditional measures of course performance, such as exams, research papers, or case studies. That research needs to relate semantic net construction with different cognitive outcomes, not just test performance. In fact, traditional test results may be the least effective variable to investigate. Tests of transfer to performance environments would be more useful dependent variables.

Another potentially important area of research with semantic networks involves changes in knowledge structures of learners as an outcome of learning. Constructing semantic networks and cognitive maps has been shown to be an accurate means for representing cognitive structure (Jonassen, 1987). That is, semantic networking helps learners to map their own cognitive structure. This affords researchers a powerful tool for verifying other learning effects. In a more recent study, Jonassen (1903) showed that building semantic networks in a course resulted in more consistent, hierarchical, and coherent knowledge structures than building expert systems in the same course. Semantic networks are a powerful knowledge analysis and integration tool that also provides means for assessing knowledge structures. Schema-based theories of learning (Rumelhart, 1980; Rumelhart & Ortony, 197.7) suggest that learning is, at least in part, a reorganization of the learner's knowledge structure. Semantic network tools provide powerful assessment tools for evaluating those changes in knowledge structure.