Using Semantic Networks for Assessment of Learners' Answers

This paper presents an assessment of learners’ knowledge about semantic networks using the Multinet paradigm. Since users of Multinet have to learn the transformation of natural language sentences in semantic networks as well as the reformulation of semantic networks in natural language two kinds of assessment are developed. In the first assessment learners build semantic networks in a graphical tool. The graphics are automatically analysed by comparison with a reference solution. In the second assessment the natural language reformulation is automatically transformed in Multinet and then compared to a reference answer. Depending on the analyses a detailed feedback is presented to the learner for the improvement of his solution.