ITSs based on the traditional architecture have problems with modularity and as aconsequence complex development and change implementation, as well as reuse of ITS components is almost impossible. Distributed technologies may be used to eliminate these drawbacks. Still, majority of the known agent based ITSs do not fully use advantages of distributed technologies. They are not open and do not enable reuse of small scale components. To solve these drawbacks three open architectures for ITS development are proposed. The main advantages of the proposed architectures are the following. All of the proposed architectures are open and consist of small-scale components that have one or very few tasks. The openness of the architectures allows creation of systems that are extendable with certain types of functionality by just adding new components and without changing existing code. Such option enables easy adaptation of the ITS’s functionality to the changes made in the course as well to any new courses. Usage of small scale components that have only one task means that these components can be reused in any system that needs this task. Additional advantages of the proposed architectures are increased modularity and decreased coupling and as a consequence complexity of interactions achieved by introduction of the organisational principles into the distributed technologies, namely multi-agent systems and service oriented architectures.
Two technologies analyzed in the chapter are similar because of their distributed nature. Therefore the architectures of systems using these technologies should be built using the same principles. Thus the lessons learned in usage of one technology can be used in building systems with another technology. The chapter shows how the lessons learned in agent based ITS development can be used in service oriented ITSs.
Agents and services have significant differences. While agents offer natural implementation of the intelligent mechanisms needed in the logical part of the system, services are easier to implement, and better known for developers. Services also support easy integration of various technologies, like various tools for the user interface of the system, video streaming tools, etc. These characteristics are important for implementation of the lower level components and the user interface. Thus agents are more suitable for some components (more deliberative components), while services are more suitable for other components (more reactive components). That is a reason why in this chapter the multi-agent and the service-oriented architectures are proposed to use as a basis to develop hybrid architecture implementing each component in the technology that fits the nature of the component. The main direction of the future work is to implement the proposed hybrid architecture and test how does it work in the practically implemented prototype of an ITS. Additionally, the possibilities to use hybrid architectures consisting of services and agents to other systems with some reactive and some deliberative components should be analysed.
- 瀏覽次數：1394
