Question-Asking vs. Model-Tracing Tutors in Dialog-Based Educational Systems

摘要

In this paper, we study the advantages and disadvantages between question-asking and model-tracing tutors. Our analysis is based on parameters acquired from some of these systems, the most important ones, such as natural language understanding mechanism, interaction-dialog managing-, and responses-natural language generation. We use tables and schemes to classify and compare these systems. The analysis results show that (a) a significant factor for the configuration of the interactions between students and tutors is the fact that the usage of dialog is much more effective than the presentation of hints, (b) most important of the two types-deep and shallow analysis-of natural language understanding is proved to be the hybrid method and (c) asking from the students to write down explanations for the steps in their proofs has a significant impact on the development of an effective learning strategy.