Difference between revisions of "Charniak and Johnson 2005"
From Cohen Courses
Jump to navigationJump to search| Line 8: | Line 8: | ||
An online version of this [[Category::Paper|paper]] can be found here: [http://aclweb.org/anthology-new/P/P05/P05-1022.pdf] | An online version of this [[Category::Paper|paper]] can be found here: [http://aclweb.org/anthology-new/P/P05/P05-1022.pdf] | ||
| + | |||
| + | == Summary == | ||
| + | |||
| + | The authors describe a method that uses [[UsesMethod::Discriminative Reranking|discriminative reranking]] for [[AddressesProblem::Parsing|parsing]]. It uses the standard Charniak parser to generate the 50 best parse trees for a given sentence. They use the 50 parse trees in a [[UsesMethod::Max Ent|MaxEnt]] reranker. The results show a small improvement over the state of the art (at the time), but is a significantly more efficient. | ||
| + | |||
| + | == Getting the <math>n</math>-best parses == | ||
| + | |||
| + | |||
| + | |||
| + | == Reranking the parses == | ||
Revision as of 23:54, 12 November 2011
This paper is a work in progress by Francis Keith
Citation
"Coarse-to-Fine n-Best Parsing and MaxEnt Discriminative Reranking", Eugene Charniak, Mark Johnson, ACL 2005
Online Version
An online version of this paper can be found here: [1]
Summary
The authors describe a method that uses discriminative reranking for parsing. It uses the standard Charniak parser to generate the 50 best parse trees for a given sentence. They use the 50 parse trees in a MaxEnt reranker. The results show a small improvement over the state of the art (at the time), but is a significantly more efficient.
