Mixtures of Hierarchical Topics with Pachinko Allo cation

Mixtures of Hierarchical Topics with Pachinko Allo cation
David Mimno - University of Massachusetts, Amherst, USA Wei Li - University of Massachusetts, Amherst, USA Andrew McCallum - University of Massachusetts, Amherst, USA
The four-level pachinko al location model (PAM) (Li & McCallum, 2006) represents correlations among topics using a DAG structure. It does not, however, represent a nested hierarchy of topics, with some topical word distributions representing the vocabulary that is shared among several more specific topics. This paper presents hierarchical PAM -- an enhancement that explicitly represents a topic hierarchy. This model can be seen as combining the advantages of hLD's topical hierarchy representation with PAM's ability to mix multiple leaves of the topic hierarchy. Experimental results show improvements in likelihood of held-out documents, as well as mutual information between automatically-discovered topics and humangenerated categories such as journals.

David Mimno - University of Massachusetts, Amherst, USA
Wei Li - University of Massachusetts, Amherst, USA
Andrew McCallum - University of Massachusetts, Amherst, USA

The four-level pachinko al location model (PAM) (Li & McCallum, 2006) represents correlations among topics using a DAG structure. It does not, however, represent a nested hierarchy of topics, with some topical word distributions representing the vocabulary that is shared among several more specific topics. This paper presents hierarchical PAM -- an enhancement that explicitly represents a topic hierarchy. This model can be seen as combining the advantages of hLD's topical hierarchy representation with PAM's ability to mix multiple leaves of the topic hierarchy. Experimental results show improvements in likelihood of held-out documents, as well as mutual information between automatically-discovered topics and humangenerated categories such as journals.