|Budget Amount *help
¥2,500,000 (Direct Cost : ¥2,500,000)
Fiscal Year 1998 : ¥1,100,000 (Direct Cost : ¥1,100,000)
Fiscal Year 1997 : ¥1,400,000 (Direct Cost : ¥1,400,000)
We have proposed multi-agent environment which assists students in discussing on a topic in a logically distinct field like physics and mathematics. The special feature of the system is as follows.
1) All students join in the discussion via each individual client computer. The student writes his/her opinion on each private window which no other student can read. The opinion is able to be transferred to a common window if he/she wants, which all participants can read.
2) An agent is assigned to each student in client site and plays a role of intelligent tutor in ITS, i. e. the ITS agent constructs a student model by using domain knowledge and teachidg knowledge.
3) Students' opinion is displayed on the windows which constitutes a node of a graph with attributes of proposition, agreement, etc. The arc of the graph consists of an arrow expressing relations between two opinions in both ends, e.g. causality, discrepancy, **stantiation, generalization, question, refutation and so on. The graph
is referredas "discussion graph"
4) Another agent, a moderator, is introduced into the server site. The moderator constructs a "group model" which is composed of the following two constituents ; one is a logical sum of all correct knowledge in all student models. The sum expresses a scope of topics which is able to be correctly inferred by students themselves. The other is the logical difference of the sum from the domain knowledge which expresses a missing knowledge of the group.
5) The moderator generates its opinion on the common window only if the discussion comes to a deadlock because of the missing knowledge.
6) The ITS agents generates a question on the common window if the pertinent student has lost the way because of the student's missing or incorrect knowledge, or if the pertinent student has written nothing on the window more than a predefined interval. If an ITS agent generates plural questions, it selects a best one from a viewpoint of tutoring. If questions of plural iTS agents coincide, they select the best one by negosiation from the instructinal view point. Thus the ITS agents have two functions of assisting the pertinent students. One is to correct misunderstanding and the other is to activate the discussion by inviting the inactive student into the discussion. Less