Discriminative Reranking for Natural Language Parsing

This paper considers approaches which rerank the output of an existing probabilistic parser. The base parser produces a set of candidate parses for each input sentence, with associated probabilities that define an initial ranking of these parses. A second model then attempts to improve upon this initial ranking, using additional features of the tree as evidence. We describe and compare two approaches to the problem: one based on Markov Random Fields, the other based on boosting approaches to reranking problems. The methods were applied to reranking output of the parser of Collins (1999) on the Wall Street Journal corpus, with a 13% relative decrease in error rate.