Deductive Error Reconstruction and Classification in a Logic Programming Framework
Deductive Error Reconstruction and Classification in a Logic Programming Framework
复制标题
逻辑编程框架中的演绎错误重构和分类
DOI：
--
复制
发表时间：
1993
期刊：
影响因子：
--
通讯作者：
H. Hoppe
中科院分区：
文献类型：
--
作者：
Sieghard Beller;H. Hoppe
虽然诊断是动态学生建模的先决条件，但我们远没有一个适用的和可转移的健全诊断方法清单。有一些证据表明，这种诊断方法的很大一部分可以在逻辑编程的框架中实现。本文的重点是使用元解释实现诊断算法在一个透明的和可重复的方式，并提高其计算效率。提出了一种具体的方法，它产生错误的假设，从故障模式在不完整的衍生树，给定一个正确的域模型和一个不正确的学生解决方案。它很容易表明，这个过程中的错误classificationwhich分离的（部分）重建学生的解决方案的错误的推理是计算成本低于经典的使用一个错误的规则库。还将讨论如何通过用于生成分类规则的学习组件来增强这种方法。
Although diagnosis is a prerequisite for dynamic student modeling, we are far from having an applicable and transferable inventory of sound diagnostic methods. There is some evidence, that a substantial part of such a diagnostic methodology could be realized in the framework of logic programming. The focus of this paper is on using meta-interpretation for implementing diagnostic algorithms in a transparent and reproducible way and also for enhancing their computational efficiency. A specific method is presented which generates bug hypotheses from failure patterns in incomplete derivation trees, given a correct domain model and an incorrect student solution. It is easily shown that this process of error classificationwhich separates the (partial) reconstruction of the student’s solution from the inference of bugs is computationally less costly than the classical use of a mal-rule library. It will also be discussed how this approach could be enhanced by a learning component for generating classification rules.