Why Intelligence?

---

First, what is the intelligence in MAS-based DLEs? Intelligence is a sort of competence for the agents to perform specified tasks while acting as human beings. The intelligence is an embedded software component that integrates software technology and complicated algorithms. The algorithms will “guide” software to perform specific tasks for human beings. Why is such intelligence necessary and important in MAS-based DLEs? To answer this question, let us start from a learning and teaching scenario in MAS-based DLEs:

Peter Orchard, a student at University, would like to take a course in Political Science, and he logs onto the portal at the University. Once he logs on, an agent [called interface agent (Lin et al., 2003) in MAS] will look at his profile, his background knowledge, and course historic records to specify a model for him and select a subject, “Politics 101” that fits his interest and background knowledge very well. So, Peter undertakes this course using learning objects provided by the learning systems. One day he is writing a report on how to set up foreign policy for his subject and would like to know other students’ opinions regarding the Sept. 11 act of terrorism. He clicks on an interface agent that communicates with other agents (they are either collaboration agents or information agents) in the distributed-learning systems, enters some keyword phrases, and identifies his preference, such as the terrorism discussion board, the chat rooms, and the necessary information agent. Then, the information agent will monitor all the course chat room and message board activities and respond to Peter’s interface agent when it observes any discussion regarding this topic. Peter can determine his action in terms of the agent’s response. For example, he can ask the information agent to collect opinions about the Sept. 11 terrorism incident from those chat rooms or message boards by authorizing the interface agent. With the collected information, Peter continues working on his report. On the next Friday, after a few days of vacation, he signs onto his education system and gets a reminding message, “Hi, Peter, you have an important note to read, and your report on ‘Politics 101’ subject is due tomorrow.” On Saturday afternoon, he is working hard to finish his report and submit it to the teacher’s agent (another interface agent) by collaboration agents. When the teacher’s agent gets the report from Peter, it will check the report and give a reasonable mark based on the criteria set by the Professor. At the same time, the teacher’s agent will send the mark to an administration agent for academic record management.

This is a typical scenario required for intelligent MAS-based DLEs. To provide such learning environments, a number of issues have to be solved. These issues are as follows:

1. How to collaborate among different agents?

2. How to secure the communication between agents?

3. How to protect the privacy of students’ information?

4. How to collect students’ behaviors and preferences data for modeling their behaviors and background knowledge or their domain knowledge (which may involve a legal issue)?

5. How to personalize the agent behavior?

6. How to make the agent adaptation to dynamic learning requirements?

7. How to negotiate the resources in the heterogonous MAS?

8. How to make a decision for choosing the suitable learning objects for the students, such as a course or subject for Peter in the above scenario?

Some of these issues are addressed in previous or upcoming chapters. In this chapter, we focus specifically on one of the most important issues, the intelligence of MAS-based DLEs.

From the above learning scenario, you have seen how important intelligence is in MAS-based DLEs. Without intelligence, it is impossible for any learning system to provide the dynamic learning environments to meet Peter’s requirements. Without intelligence, it is impossible to reach personalization and adaptation for DLEs; it is also difficult to build efficient models for student behavior and background knowledge. Therefore, the MAS-based DLEs must possess intelligent competence and provide intelligent support.

The imminent issue now is the development of intelligence by applying AI techniques. This will be addressed in detail in the following three sections.

---