Discussion and Directions for Future Research

---

The benefits of using Gamma as a specification language for agent systems is not limited to the method it provides for accurate agent system design. Because Gamma is a functional programming language, it enables us to reason about the properties of the specified system (Le Metayer, 1996; Hankin et al., 1993a,b; Reynolds, 1996; Ma et al., 1994; Ma & Orgun, 1995; Lin et al., 1997, 1998; Lin & Chen, 1998). In Reynolds (1996), Ma et al. (1994), Ma & Orgun (1995), and Lin et al. (1997), temporal logic is used to define the semantics of the Gamma language. Temporal logic semantics has also been used to facilitate a program-transformation method that provides a way to implement the Gamma language. Other studies of the semantics of the Gamma language include the study of categorical semantics (Lin & Chen, 1998; Lin et al., 1998). In Le Metayer (1996), the Gamma language is used to reason about the type of software architecture. More research will be needed to determine how these semantic models can be used to address the design issues of the e-learning environment.

We have examined the feasibility of using the Gamma language to specify agent systems. Our study was carried out in a generic e-learning context. More specific studies should be done to learn how the Gamma language can work with existing e-learning models, for example, the Learner Model (Gavrilova et al., 1999), so that area specific problems can be identified and addressed.

---