シラバス-専門用語の相互クラスタリングを用いたカリキュラム分析システムの改善

Comprehending the features of curricula provided by various higher education institutions is significant in designing and evaluating a curriculum. To facilitate comprehension of the curricula's features, Curriculum Analyzing System has been developed by Nozawa et al. utilizing document-clustering of syllabus data. However, speeding up and improvement of interactivity in the clustering procedure remained to be accomplished. In this article, a co-clustering method based on graph partitioning is introduced in order to improve the curriculum analyzing system. We show that recursive application of co-clustering to syllabus-term data makes the system more efficient and enables the users to understand the curricula's features more interactively.