Why Use Intelligent Agents for Adaptive E-Learning?

---

The main objective of e-learning systems is to enable individually subscribedto learning services to be delivered to their associated users whenever they request them, and wherever the users are, in a customized form that matches their profile. Thus, intelligent mobile agents have been introduced to provide this kind of dynamic service provisioning and management. The agents’ technology has several advantages for implementing new services on distributed systems. In fact, this technology enables these systems to distribute the functionalities in small, reproducible, and distributed software entities. It also allows for a clear and easy separation between their internal, private knowledge, and their interface toward the external world and other agents.

The e-learning services provisioning and management fits well for exploiting the agents’ properties.

• Agents’ autonomy:

  • Allows for making decisions on service access, the interfaces’ configuration, and service provisioning without human assistance.

  • Allows for automating the control and management tasks and, hence, reducing the operator’s workload.

  • Allows for automating the service deployment and provision, thus reducing the effort and time required for the installation and the maintenance of services.

• Agents’ intelligence:

  • Allows for the dynamic customization and configuration of services. The agents can learn and adapt to the preferences of their users and detect and update old versions of services.

  • Allows for the service intelligence to be downloaded dynamically from the providers and for collaboration between different providers.

• Agents’ mobility:

  • Supports the dynamic topological of service provisioning.

  • Enables e-learning services to be provided instantly and to be customized directly at the locations where services are needed.

  • Enables dynamic provision of customized services by downloading service agents from the service-provider system to the network nodes or user terminals.

• Agents’ sociability:

  • Offers the potential to distribute service-related processing, and also offers a mechanism for the nodes in different networks to cooperate in order to provide a service to the user.

  • Allows for negotiation for service features.

  • Provides multiservices interaction and coordination.

  • Allows for the asynchronous and cooperative processing of tasks.

  • The agents’ technology fits well for e-learning, because it supports the following requirements:

• Dynamic scalability:

  • Mobile intelligent agents (MIAs) support huge distributed systems such as the Internet. In fact, each service is modeled as a collection of agents, each agent occupying different places at different times, because it can move from one place to another.

  • MIAs support on-demand provisioning of services. In fact, when servers are implemented with MIAs, the agents’ mobility allows them to deploy new replicas when the demand arises or to migrate to the location where the demand is high.

  • MIAs enable the provision of flexible solutions, in which services are partitioned into mobile service agents achieving multiple functions that can be spread across the network.

• Distribution of services:

  • MIAs fit well for modeling the ideal situation for a mobile user, for mobile agents can provide the ubiquitous availability of applications, data files, and user profiles by using the concepts of mobility and cloning.

  • MIAs enable control tasks to be performed in a distributed manner by using the concept of remote programming (Fuggetta et al., 1998) instead of the client/server programming concept used currently in most Web-based learning systems.

  • The possibility of bringing control or management agents as close as possible to the resources allows for a more decentralized realization of service control and management software than could be achieved otherwise.

• Reduction of traffic:

  • MIAs decrease pressure on centralized network-management systems and network bandwidth by using spatial distribution and temporal distribution.

  • MIAs’ autonomous and asynchronous operation reduces the requirements regarding traffic load and the availability of the underlying networks

• Independence regarding failures:

  • MIAs reduce the influence of signaling network faults during service processing, because once a service agent has migrated, the processing will be performed locally.

  • The agents’ migration to the required data reduces dependence regarding network availability, so more robustness is achieved in the distributed system.

Finally, agent technology offers a number of very interesting advantages, but it should not be seen as the only solution for all Web services. Rather, it should be seen as a technology that can resolve some problems. Furthermore, we have to consider some of MIAs’ disadvantages:

• Require a specific run-time environment (agent execution environment) to be present in all nodes to be visited.

• Create a security problem. The platforms have to be protected from malicious agents and vice versa.

• May increase the network load in some situations. One of the mobile agents’ goals is to reduce the network traffic, but it does not seem useful for every agent to migrate in every situation; doing so would probably increase the network traffic. Therefore, new strategies have to be developed to establish under which circumstances an agent will migrate.

• Do not provide location transparency. Each agent must be aware of the location to be visited.

The MIAs technology is relatively new, and its suitability for solving Webbased services problems needs to be demonstrated by implementing more systems using this technology.

---