Paper: Cross-Task Knowledge-Constrained Self Training

ACL ID D08-1071
Title Cross-Task Knowledge-Constrained Self Training
Venue Conference on Empirical Methods in Natural Language Processing
Session Main Conference
Year 2008

We present an algorithmic framework for learning multiple related tasks. Our frame- work exploits a form of prior knowledge that relates the output spaces of these tasks. We present PAC learning results that analyze the conditions under which such learning is pos- sible. We present results on learning a shal- low parser and named-entity recognition sys- tem that exploits our framework, showing con- sistent improvements over baseline methods.