id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_wqab4ttubbdubhqub6dpdmptmq	Ryan J. Gallagher	Anchored Correlation Explanation: Topic Modeling with Minimal Domain Knowledge	2017	14	.pdf	application/pdf	8961	1194	67	(CorEx), an alternative approach to topic modeling that does not assume an underlying generative model, and instead learns maximally domain knowledge can be flexibly incorporated within CorEx through anchor words, allowing topic separability and representation to 1Open source, documented code for the CorEx topic model may flexibly incorporate word-level domain knowledge within the CorEx topic model. can be naturally integrated into CorEx through "anchor words" and the information bottleneck. treat anchor words as fuzzy logic markers and embed them into the topic model in a semi-supervised anchor one word to multiple topics, allowing CorEx We compare CorEx to LDA in terms of topic coherence, document classification, and document clustering across three datasets. Figure 3: Comparison of anchored CorEx to other semisupervised topic models in terms of document clustering CorEx topic model with that label's anchor words of the CorEx topic model to LDA, it does have some	./cache/work_wqab4ttubbdubhqub6dpdmptmq.pdf	./txt/work_wqab4ttubbdubhqub6dpdmptmq.txt