A Synthetic Agent for Mentoring Novice Programmers Within a Desktop Computer Environment

Abstract

This research proposes the hypothesis that a synthetic animated agent can provide effective mentoring support for the novice in overcoming problems and pitfalls when learning a programming language for the very first time. Computer programming is a skills-based activity that involves problem solving within the constraints imposed by a computer environment. Numerous authorities have observed that novice programmers make the same mistakes and encounter the same problems when first programming language. The learner errors are usually from a fixed set of misconceptions that are easily corrected by experience and with simple guidance. This research investigates the viability of a synthetic animated agent to provide effective guidance to mentor novice programmers.

The primary pedagogical model that the mentor agent will use is based on the Scaffolding model, where a tutor provides temporary support to the learner and removes the support as it is no longer needed. The support will take many forms e.g. explanations, examples, direction, etc. but requires the learner to be active participants in producing work of their own. For the agent mentoring systems it is planned that the learner will continue to be lectured and tutored as normal with the mentoring agent to provide support in the practical, skills based activity of writing code. The agent will monitor the learner’s development environment as they proceed with writing code, the animated agent would then impart guidance when the learner generates errors or unexpected results.

An innovative agent architecture is then proposed based on an analysis of the problem domain that will consist of a Beliefs-Desires-Intentions agent model, to control the animated agent interaction and interface with the programming environment, combined with a Case-Based Reasoning engine to process deep domain knowledge and guide the overall mentoring strategy.

The mentoring agent will operate alongside the development environment for the programming language, for this research Visual Basic and within the Microsoft Windows environment. Experimentation with a working agent system is planned to begin in the autumn of 2005 with cohorts of undergraduate programming novices.