Fig 1.
Hierarchy among topics: A circle represents a topic, dashed rectangle represents a path, each layer represents a level, and rectangular nodes represent the candidate branch that might be added by the nCRP.
Table 1.
Summary of qualitative characteristics.
Table 2.
Running time complexities.
Table 3.
Selected topics related to “Rivers” and “Banks/Financial” in the Wikipedia dataset.
Fig 2.
Topic hierarchy estimated using GhLDA.
Fig 3.
Low-dimensional representation of words and their topic assignments using GLDA.
Fig 4.
Low-dimensional representation of words and their topic assignments using GhLDA.
Fig 5.
Partial depiction of the topic hierarchy estimated using hLDA.
Table 4.
Topic coherence.
Table 5.
Predictive held-out likelihood.