Topic Models for Corpus-centric Knowledge Generalization
Benjamin Van Durme, Daniel Gildea · UR Research (University of Rochester) · 2009
Many of the previous efforts in generalizing over knowledge extracted from text have relied on the use of manually created word sense hierarchies, such as WordNet. We present initial results on generalizing over textually derived knowledge, through the use of the LDA topic model framework, as the first step towards automatically building corpus specific ontologies. This work was funded by a 2008 Provost’s Multidisciplinary Award from the University of Rochester, and NSF Many of the previous efforts in generalizing knowledge extracted from text (e.g., Suchanek et al. (2007), Banko and Etzioni (2007), Pas¸ca (2008), and Van Durme et al. (2009)) have relied on the use of manually created word sense hierarchies, such as WordNet. Unfortunately, as these hierarchies are constructed based on the intuitions of lexicographers or knowledge engineers, rather than