Moving towards a Fully Automatic Knowledge Assessment Tool

Abstract

Information about a studentâ??s level or state of knowledge is a key aspect for efficient, personalized learning activities. E-learning systems gain such information in two ways: directly by examining usersâ?? self-assessment and administering predefined tests and indirectly by making inferences on observed user behaviors. However, most of the current solution approaches either demand excessive manpower or lack required reliability. To overcome these problems, we have developed the e-Examiner, an assessment tool that supports the assessment process by creating automatically test items, assessing studentsâ?? answers and providing feedback. In this paper, we firstly give an overview about a variety of computer-assisted and computer-based assessment systems and methods that support formative assessment activities. Secondly, we introduce the overall concept and architecture of the e-Examiner. Thirdly, we outline implementation details and evaluation results of our prototype implementation. Our solution approach is based on the set of statistical similarity measures defined by the ROUGE toolset for automatic summary evaluation.