Research And Implementation For The Personalized Collaborative Learning System Based On Multi-agent Technology

Computer supported collaborative learning is a reform of traditional education. It helps to make the process of education beyond the shackle of time and space and therefore gets great attention from researchers. This paper is aimed at exploring two issues: the personal learning in the condition of distance collaborative learning in which the technology of agent is introduced, and new methods for Allied Coordinating Learning. Meanwhile, we also intend to establish a personal distance collaborative learning enviroment based on the Multi-Agent technology. The main tasks involved in the paper are as follows:1. A model of Multi-Agent Collaborative Learning System based on the knowledge level of learners is presented to solve the problem in the existent learning systems, such as the universal out-of-order and confusion, run short of individuation and effects of population interactions. The model takes into account the personal properties like learners' knowledge, cognizing characters and personal interests, and creates a storied structure by defining learners' knowledge level.2. With regard to characteristics of Collaborative Learning System based on the Multi-Agent technology, we give a Simple-Agent capacity describing language. It combinates the important indexes of performance: input/output constrains, quality, duration and cost to distinguish the agents' ability. Describing Agent's capacity is conducive to a better understanding between agents and a more appropriate match between tasks and capacity. It can also realize capacity-based reasoning, programming and learning, whereas helps users to use the system more easily.3. A grouping strategy based on learners' learning goals is also provided, which is intended to meet need of the system. This new Allied Strategy of behavior criteria suits agent's allied solution in a dynamic opening situation and improves the efficiency of searching solutions in Multi-Agent Learning System. With help of this strategy, we can construct optimal solution plan for common learning goals and avoid effectively the allied relay harm to general efficiency, which maintains the system's stable and helps it gain a general optimal solution.4. According to the main problem in existent Multi-Agent Collaborative Learning Systems, i.e., task distribution between agents, we give a kind of arithmetic for distribution learning tasks, which is based on map. The paper uses structure of map to indicate state space of problems and to change the process of task distribution to a process of map. Furthermore, on the basis of this arithmetic, we suggest a method based on model of state space search to realize personal task distribution based on different agents' knowledge set.5. Picking the idea of object-oriented analysis and design, we design and develop a personal Multi-Agent Collaborative Learning archetypal system based on learners' knowledge level.