You are here

Hybrid intelligent tutoring system architecture

15 January, 2016 - 09:50

As concluded above, agents are more suitable to implement some components than services and vice versa for other components. To benefit from the advantages of both technologies hybrid architecture is proposed. It preserves the approach used in the previously described architectures in two senses. Firstly, each module from the traditional ITS architecture is implemented as a set of distributed components. Secondly, the architecture consists of two levels. Still the implementation differs from the previously described architectures. The higher level consists of agents that implement the logical or deliberative part of the system. The agents correspond to the higher level holons of the multi-agent architecture. The lower level of the architecture is mainly implemented as services that carry out simple reactive behaviour upon request like retrieving data from repositories. Agents have the same functions as heads of the holons in the multi-agent architecture – they make all intelligent choices to carry out the tutoring process and use services for particular tasks. Services of the hybrid architecture are used instead of the body agents of the multi-agent architecture. It allows keeping the advantages of both previously described architectures, like openness to new functionality and high modularity. The proposed architecture is presented in the following subsections by specifying agents of each module as well as services used by these agents and, as a consequence, included in the corresponding modules.