2006 EMNLP EMNLP 2006

Solving the Problem of Cascading Errors: Approximate Bayesian Inference for Linguistic Annotation Pipelines