Analysis of Computer Science Textbooks by Topic Modeling and Dynamic Time Warping

Taisuke Kawamata, Yoshitatsu Matsuda, Takayuki Sekiya, Kazunori Yamaguchi

Abstract: The design of appropriate curricula in the rapidly expanding field of computer science (CS) is one of the most important issues in higher education, and there are many features to be considered. To provide useful information for developing appropriate curricula, we investigated the actual textbooks traditionally used in CS by applying the latent Dirichlet allocation to actual textbooks. In this paper, we focused on the topic progression of the textbooks and showed the results of clustering analysis with sequence clustering. We discovered that the structure of CS textbooks can be explained by five or six topics, and that dynamic time warping (DTW) can classify knowledge areas for CS more appropriately than the LDA-based clustering without DTW.