Implementation patterns for supporting learning and group interactions

Abstract

This thesis covers research on group learning by using a computer as the medium. The computer software provides the basic blending of the students contributions augmented by the effects generated for the specific learning domain by a system of agents to guide the process of the students learning. The research is based on the approach that the computer as a medium is not an end point of the interaction. The development of agents in based on Human-Computer-Human interaction or HCH. HCH is about removing the idea that the role of the computer is that of an intelligent agent and reducing its role to that of a mixer, with the ability to insert adaptive electronic (software) components that add extra effects and depth to the product of the human-human interactions. For the computer to achieve this support, it must be able to analyse the input from the individuals and the group as a whole. Experiments have been conducted on groups working face to face, and then on groups using software developed for the research. Patterns of interaction and learning have been extracted from the logs and files of these group sessions. Also a pattern language has been developed by which to describe these patterns, so that the agent support needed to analyse and respond appropriately to each pattern can be developed. The research has led to the derivation of a structure that encompasses all the types of support required, and provides the format for implementing each type of support. The main difficulty in this work is the limited ability of computers to analyse human thoughts through their actions. However progress is made in analysing the level of approach by students to a range of learning concepts. The research identified the separate patterns that contribute to learning agents development and form a language of learning processes, and the agents derived from these patterns could in future be linked into a multi-agent system to support learning.