This document proposes using collaborative knowledge bases like Wikipedia to help understand topics discovered by topic modeling algorithms. It describes how Wikipedia pages can represent topics by characterizing them as distributions over words in the pages. An experiment matches topics from LDA modeling of journal abstracts to Wikipedia categories, achieving a match rate of 50.8% for primary categories and 29.5% for primary categories alone using only word distributions in Wikipedia pages. Next steps include modeling dependent topic structures and combining heuristics like Wikipedia with generative models.