Spherical Topic Models
Joseph Reisinger, Austin Waters, Bryan Silverthorn, Raymond J. Mooney · 2010
We introduce the Spherical Admixture Model (SAM), a Bayesian topic model for arbitrary `2 normalized data. SAM maintains the same hi-erarchical structure as Latent Dirichlet Alloca-tion (LDA), but models documents as points on a high-dimensional spherical manifold, allowing a natural likelihood parameterization in terms of cosine distance. Furthermore, SAM can model word absence/presence at the document level, and unlike previous models can assign explicit negative weight to topic terms. Performance is evaluated empirically, both through human rat-ings of topic quality and through diverse classi-fication tasks from natural language processing and computer vision. In these experiments, SAM consistently outperforms existing models. 1.