How to Generate Exercise Questions for Web-Based Investigative Learning

Abstract

In recent years, conducting investigative learning on the Web (Web-based investigative learning) has become popular, in which learners could investigate various questions with a huge number and variety of Web resources. This involves navigating Web resources/pages and constructing knowledge. In an elaborate investigation about an initial question, learners are expected to widen and deepen the question, which requires them to identify related questions to be further investigated. This corresponds to expanding the initial question into related ones as sub-questions. Such question expansion allows learners to create their own learning scenario. However, it is not easy for learners to create it in concurrence with navigation and knowledge construction process. In our previous work, we have proposed a model of Web-based investigative learning, and developed cognitive tool called interactive Learning Scenario Builder (iLSB for short). The results of the case study we conducted suggest that iLSB could significantly promote question expansion. The issue addressed in this paper is how to develop skills in question expansion for Web-based investigative learning with iLSB. Towards this issue, it is necessary for learners to practise so that they could expand an initial question wider and deeper. In this paper, we propose a method for generating exercise questions, which extracts keywords representing candidates of sub-questions to be expanded from an initial question by means of Linked Open Data and Word2vec. This paper also reports the case studies whose purpose were to evaluate the appropriateness and effectiveness of exercise questions generated by the proposed method. The results suggest that exercise questions are appropriate and effectively work particularly for novice learners to conduct question expansion.