[1J2-J-6] Web mining 1

[1J2-J-6-02] Visualization and Analysis of the Hierarchical Correlation among Words and Documents in Hyperbolic Space

〇Daiki Hashimoto¹, Hori Koichi¹ (1. University of Tokyo)

Keywords:Poincare Embedding, Document Mining

In areas such as Aerospace Engineering where many areas come together to form one complicated area of study, it could be difficult to grasp its entire structure especially for students who has just began learning it. We address this problem by visualizing the hierarchical relationship of different papers in such areas in hyperbolic space to give an overview of the area and suggest which papers to read in what kind of order according to the users interest. We also discoverd ways to find new insights in relationship among different topics by using Wasserstein Metrics with the created visualization.