Parsing (State of the art)
From ACLWiki
(Difference between revisions)
Slavpetrov (Talk | contribs) |
|||
| Line 1: | Line 1: | ||
| − | + | * '''Performance measure:''' PARSEVAL - the [http://nlp.cs.nyu.edu/evalb/ evalb] program | |
| − | PARSEVAL | + | * '''Training data:''' sections 2-22 of Wall Street Journal corpus |
| + | * '''Testing data:''' section 23 of Wall Street Journal corpus | ||
| − | |||
| − | |||
| − | |||
| − | { | + | {| border="1" cellpadding="5" cellspacing="1" width="100%" |
| − | + | |- | |
| − | | Johnson & Charniak's Parser | + | ! System name |
| + | ! Short description | ||
| + | ! Main publications | ||
| + | ! Software | ||
| + | ! Results (PARSEVAL) | ||
| + | ! Comments | ||
| + | |- | ||
| + | | Johnson & Charniak's Parser | ||
| + | | Lexicalized N-Best PCFG + Discriminative re-reanking | ||
| + | | Mark Johnson and Eugene Charniak, ACL 2005 [http://acl.ldc.upenn.edu/P/P05/P05-1022.pdf] | ||
| + | | Available [http://www.cog.brown.edu/~mj/Software.htm] | ||
| + | | 91.4% | ||
| + | | works well also on Brown | ||
|- | |- | ||
| − | | Collins' Parser | + | | Collins' Parser |
| + | | Lexicalized PCFG | ||
| + | | Collins' Thesis [http://people.csail.mit.edu/mcollins/papers/thesis.ps], Bikel's Thesis [http://www.cis.upenn.edu/~dbikel/papers/thesis.pdf] | ||
| + | | Dan Bikel's Implementation [http://www.cis.upenn.edu/~dbikel/software.html] | ||
| + | | ??? | ||
| + | | | ||
|- | |- | ||
| − | | Berkeley Parser | + | | Berkeley Parser |
| − | + | | Automatically induced PCFG | |
| + | | Petrov, Barrett, Thibaux and Klein, ACL 2006 [http://www.eecs.berkeley.edu/~petrov/data/acl06.pdf], Petrov and Klein, NAACL 2007 [http://www.eecs.berkeley.edu/~petrov/data/naacl07.pdf] | ||
| + | | Available [http://nlp.cs.berkeley.edu/Main.html#Parsing] | ||
| + | | 90.1% | ||
| + | | works well also for Chinese and German | ||
|- | |- | ||
|} | |} | ||
Revision as of 13:21, 27 June 2007
- Performance measure: PARSEVAL - the evalb program
- Training data: sections 2-22 of Wall Street Journal corpus
- Testing data: section 23 of Wall Street Journal corpus
| System name | Short description | Main publications | Software | Results (PARSEVAL) | Comments |
|---|---|---|---|---|---|
| Johnson & Charniak's Parser | Lexicalized N-Best PCFG + Discriminative re-reanking | Mark Johnson and Eugene Charniak, ACL 2005 [1] | Available [2] | 91.4% | works well also on Brown |
| Collins' Parser | Lexicalized PCFG | Collins' Thesis [3], Bikel's Thesis [4] | Dan Bikel's Implementation [5] | ??? | |
| Berkeley Parser | Automatically induced PCFG | Petrov, Barrett, Thibaux and Klein, ACL 2006 [6], Petrov and Klein, NAACL 2007 [7] | Available [8] | 90.1% | works well also for Chinese and German |