id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_hhljuimcpfdjrp7qy2oosk7bry	Michael J. Paul	Sprite: Generalizing Topic Models with Structured Priors	2015	15	.pdf	application/pdf	10143	1158	66	SPRITE: Generalizing Topic Models with Structured Priors 2 Topic Modeling with Structured Priors are real-valued vectors of length equal to the vocabulary size V (for priors over word distributions) or length equal to the number of topics K By modeling the priors as combinations of components that are shared across all topics, we can learn interesting connections between SPRITE: Structured PRIor Topic modEls. To illustrate the role that components can play, Figure 2: Example graph structures describing possible relations between components (middle row) and topics or documents Edges correspond to non-zero values for α or β (the component coefficients defining priors over the document Table 1: Topic models with Dirichlet priors that are generalized by SPRITE. Each supertopic is associated with a Dirichlet prior over subtopic distributions, where subtopics are the low level topics that are associated with word parameters φ.	./cache/work_hhljuimcpfdjrp7qy2oosk7bry.pdf	./txt/work_hhljuimcpfdjrp7qy2oosk7bry.txt