Paper: Maximum Entropy Modeling In Sparse Semantic Tagging

ACL ID N04-2003
Title Maximum Entropy Modeling In Sparse Semantic Tagging
Venue Human Language Technologies
Session Student Session
Year 2004

In this work, we are concerned with a coarse grained semantic analysis over sparse data, which labels all nouns with a set of semantic categories. To get the benefit of unlabeled data, we propose a bootstrapping framework with Maximum En- tropy modeling (MaxEnt) as the statistical learn- ing component. During the iterative tagging pro- cess, unlabeled data is used not only for better statistical estimation, but also as a medium to in- tegrate non-statistical knowledge into the model training. Two main issues are discussed in this paper. First, Association Rule principles are sug- gested to guide MaxEnt feature selections. Sec- ond, to guarantee the convergence of the boot- strapping process, three adjusting strategies are proposed to soft tag unlabeled data.