Search Descriptions

Main Topics

Search Publications


author

title

other

year

Main.HomePage History

Hide minor edits - Show changes to output

March 29, 2013, at 10:40 AM by Philipp Koehn -
Deleted line 2:
(:search-publications-sidebar:)
March 29, 2013, at 10:39 AM by Philipp Koehn -
Added lines 3-4:
(:search-publications-sidebar:)
Deleted line 8:
(:search-publications-sidebar:)
March 29, 2013, at 10:39 AM by Philipp Koehn -
Added lines 7-8:
(:search-publications-sidebar:)
Changed line 23 from:
The topic structure may be [[edited -> Meta/AddBibTeX]], by adding new topics, re-arranging the hierarchy, but there is currently no support for renaming topics via the Wiki. Some of the larger topics should be broken up.
to:
The topic structure may be [[edited -> Meta/AddBibTeX]], by adding new topics, re-arranging the hierarchy, but there is currently no support for renaming topics via the Wiki. Some of the larger topics should be broken up.
March 28, 2013, at 11:28 PM by Philipp Koehn -
Changed lines 1-3 from:
!Welcome

This Wiki covers all publications on all topics in the field of statistical machine translation. (:stats:)
to:
'''This Wiki covers all publications on all topics''' in the field of statistical machine translation. (:stats:)
March 28, 2013, at 11:27 PM by Philipp Koehn -
Changed line 18 from:
Citations may be added by pasting their BibTex entries into a [[text box -> Meta/AddBibTeX]]. Please ensure that the publication is not already included under a different BibTex label. If the label is the same, then duplicates are automatically detected. If an online source for the pdf file of the paper is known, include it with BibTex field @@url@@. BibTex entries may be edited at a later stage.
to:
Citations may be added by pasting their BibTeX entries into a [[text box -> Meta/AddBibTeX]]. Please ensure that the publication is not already included under a different BibTeX label. If the label is the same, then duplicates are automatically detected. If an online source for the pdf file of the paper is known, include it with BibTeX field @@url@@. BibTeX entries may be edited at a later stage.
March 28, 2013, at 11:25 PM by Philipp Koehn -
Changed line 10 from:
Most citations have links to pdf versions of the publications online, but if one does not, you can help. Follow the link to Google Scholar, or do other searches and then add the url to the BibTeX entry.
to:
Most citations have links to pdf versions of the publications online, but if one does not, you can help. Follow the link to Google Scholar, or do other searches, and then add the url to the BibTeX entry.
March 28, 2013, at 11:25 PM by Philipp Koehn -
Added lines 9-11:
!! Find Publications Online
Most citations have links to pdf versions of the publications online, but if one does not, you can help. Follow the link to Google Scholar, or do other searches and then add the url to the BibTeX entry.

Deleted lines 23-29:

!Beta

More to do:
* conferences: EAMT 2008-2010, 2006 and earlier
* automatic finding of pdfs for papers
* cache pdfs locally, allow download individually or all of them (per topic?) in a linked pdf
March 26, 2013, at 10:39 PM by Philipp Koehn -
Deleted lines 7-8:

Help out by...
March 26, 2013, at 10:39 PM by Philipp Koehn -
Changed line 11 from:
!! Improving the Descriptions
to:
!! Improve the Descriptions
Changed line 14 from:
!! Adding Publications
to:
!! Add Publications
Changed line 21 from:
!! Adding Topics
to:
!! Add Topics
March 26, 2013, at 10:38 PM by Philipp Koehn -
Changed lines 27-28 from:
* conferences: IWSLT 2011
* contact key people for comments
to:
* conferences: EAMT 2008-2010, 2006 and earlier
Changed lines 29-30 from:
* cache pdfs locally, allow download individually or all of them (per topic?) in a linked pdf
* paper stats, trending topics
to:
* cache pdfs locally, allow download individually or all of them (per topic?) in a linked pdf
March 26, 2013, at 08:59 PM by Philipp Koehn -
Added lines 8-9:

Help out by...
March 26, 2013, at 08:58 PM by Philipp Koehn -
Changed lines 3-5 from:
This Wiki covers all publications on all topics in the field of statistical machine translation.

(:stats:)
to:
This Wiki covers all publications on all topics in the field of statistical machine translation. (:stats:)
March 26, 2013, at 08:58 PM by Philipp Koehn -
Deleted lines 7-11:

Please contribute by
* improving the descriptions
* [[adding citations -> Meta/AddBibTeX]]
* [[adding topics -> Meta/Topics]]
March 26, 2013, at 08:57 PM by Philipp Koehn -
Added line 6:
March 26, 2013, at 08:57 PM by Philipp Koehn -
Added line 5:
(:stats:)
March 26, 2013, at 08:45 PM by Philipp Koehn -
Changed line 4 from:
(:stats:)
to:
March 26, 2013, at 08:41 PM by Philipp Koehn -
March 26, 2013, at 08:39 PM by Philipp Koehn -
March 26, 2013, at 08:37 PM by Philipp Koehn -
Added line 5:
(:trending-topics:)
March 26, 2013, at 06:02 PM by Philipp Koehn -
Deleted line 3:
March 26, 2013, at 06:00 PM by Philipp Koehn -
Changed line 3 from:
The goal of this Wiki is to cover all publications on all topics in the field of statistical machine translation.
to:
This Wiki covers all publications on all topics in the field of statistical machine translation.
March 26, 2013, at 05:52 PM by Philipp Koehn -
Added lines 4-5:

(:stats:)
December 17, 2012, at 05:32 PM by 129.215.24.18 -
Changed line 1 from:
!Welcomee
to:
!Welcome
December 17, 2012, at 04:18 PM by 129.215.24.18 -
Changed line 1 from:
!Welcome
to:
!Welcomee
December 17, 2012, at 04:07 PM by 129.215.24.18 -
Changed lines 1-32 from:
qjRUDF  <a href="http://sgheyutcqbyl.com/">sgheyutcqbyl</a>, [url=http://fzkcohltfbng.com/]fzkcohltfbng[/url], [link=http://rxfhicyiprfx.com/]rxfhicyiprfx[/link], http://yfzqjrjduwjh.com/
to:
!Welcome

The goal of this Wiki is to cover all publications on all topics in the field of statistical machine translation.

Please contribute by
* improving the descriptions
* [[adding citations -> Meta/AddBibTeX]]
* [[adding topics -> Meta/Topics]]

Stay alert by [[subscribing -> Meta/Subscription]] to updates to the Wiki.

!! Improving the Descriptions
Currently, the text for most topics is quite rough, since it was pieced together sentence by sentence whenever a new paper appeared. A better structure would be introducing the main concepts alongside the key papers before discussing incremental improvements. However, do not remove papers because you deem them not good enough. The goal is to cite all papers and summarize their key contribution. Please read the [[Meta/style guide]].

!! Adding Publications
Adding publications involves three steps: [[adding BibTex entries -> Meta/AddBibTeX]], [[assigning them to topics -> Meta/AssignPublications]], and describing them in the topic description.

Citations may be added by pasting their BibTex entries into a [[text box -> Meta/AddBibTeX]]. Please ensure that the publication is not already included under a different BibTex label. If the label is the same, then duplicates are automatically detected. If an online source for the pdf file of the paper is known, include it with BibTex field @@url@@. BibTex entries may be edited at a later stage.

[[Assigning publications to topics -> Meta/AssignPublications]] adds them to the section "New Publications", which is a first step and at least makes readers of the topic description aware of them.

!! Adding Topics
The topic structure may be [[edited -> Meta/AddBibTeX]], by adding new topics, re-arranging the hierarchy, but there is currently no support for renaming topics via the Wiki. Some of the larger topics should be broken up.

!Beta

More to do:
* conferences: IWSLT 2011
* contact key people for comments
* automatic finding of pdfs for papers
* cache pdfs locally, allow download individually or all of them (per topic?) in a linked pdf
* paper stats, trending topics
November 20, 2012, at 02:39 PM by tbqyrxldtms - JjsPJKPyfUsPbDZucMO
Changed lines 1-32 from:
!Welcome

The goal of this Wiki is to cover all publications on all topics in the field of statistical machine translation.

Please contribute by
* improving the descriptions
* [[adding citations -> Meta/AddBibTeX]]
* [[adding topics -> Meta/Topics]]

Stay alert by [[subscribing -> Meta/Subscription]] to updates to the Wiki.

!! Improving the Descriptions
Currently, the text for most topics is quite rough, since it was pieced together sentence by sentence whenever a new paper appeared. A better structure would be introducing the main concepts alongside the key papers before discussing incremental improvements. However, do not remove papers because you deem them not good enough. The goal is to cite all papers and summarize their key contribution. Please read the [[Meta/style guide]].

!! Adding Publications
Adding publications involves three steps: [[adding BibTex entries -> Meta/AddBibTeX]], [[assigning them to topics -> Meta/AssignPublications]], and describing them in the topic description.

Citations may be added by pasting their BibTex entries into a [[text box -> Meta/AddBibTeX]]. Please ensure that the publication is not already included under a different BibTex label. If the label is the same, then duplicates are automatically detected. If an online source for the pdf file of the paper is known, include it with BibTex field @@url@@. BibTex entries may be edited at a later stage.

[[Assigning publications to topics -> Meta/AssignPublications]] adds them to the section "New Publications", which is a first step and at least makes readers of the topic description aware of them.

!! Adding Topics
The topic structure may be [[edited -> Meta/AddBibTeX]], by adding new topics, re-arranging the hierarchy, but there is currently no support for renaming topics via the Wiki. Some of the larger topics should be broken up.

!Beta

More to do:
* conferences: IWSLT 2011
* contact key people for comments
* automatic finding of pdfs for papers
* cache pdfs locally, allow download individually or all of them (per topic?) in a linked pdf
* paper stats, trending topics
to:
qjRUDF  <a href="http://sgheyutcqbyl.com/">sgheyutcqbyl</a>, [url=http://fzkcohltfbng.com/]fzkcohltfbng[/url], [link=http://rxfhicyiprfx.com/]rxfhicyiprfx[/link], http://yfzqjrjduwjh.com/
May 04, 2012, at 03:10 AM by Philipp Koehn -
Added lines 9-10:

Stay alert by [[subscribing -> Meta/Subscription]] to updates to the Wiki.
May 04, 2012, at 03:07 AM by Philipp Koehn -
Changed line 11 from:
Currently, the text for most topics is quite rough, since it was pieced together sentence by sentence whenever a new paper appeared. A better structure would be introducing the main concepts alongside the key papers would be helpful before discussing incremental improvements. However, do not remove papers because you deem them not good enough. The goal is to cite all papers and summarize their key contribution. Please read the [[Meta/style guide]].
to:
Currently, the text for most topics is quite rough, since it was pieced together sentence by sentence whenever a new paper appeared. A better structure would be introducing the main concepts alongside the key papers before discussing incremental improvements. However, do not remove papers because you deem them not good enough. The goal is to cite all papers and summarize their key contribution. Please read the [[Meta/style guide]].
May 04, 2012, at 03:07 AM by Philipp Koehn -
Changed line 16 from:
Citations may be added by pasting their BibTex entries into a [[text box -> Meta/AddBibTeX]]. Please ensure that the publication is not already included under a different BibTex label. If the label is the same, then duplicates are automatically detected. If an online source for the pdf file of the paper is known, include it with BibTex field @@url@@.
to:
Citations may be added by pasting their BibTex entries into a [[text box -> Meta/AddBibTeX]]. Please ensure that the publication is not already included under a different BibTex label. If the label is the same, then duplicates are automatically detected. If an online source for the pdf file of the paper is known, include it with BibTex field @@url@@. BibTex entries may be edited at a later stage.
May 04, 2012, at 03:06 AM by Philipp Koehn -
Changed lines 13-15 from:
!! Adding citations
to:
!! Adding Publications
Adding publications involves three steps: [[adding BibTex entries -> Meta/AddBibTeX]], [[assigning them to topics -> Meta/AssignPublications]], and describing them in the topic description.

Changed lines 18-20 from:
!! Topics
to:
[[Assigning publications to topics -> Meta/AssignPublications]] adds them to the section "New Publications", which is a first step and at least makes readers of the topic description aware of them.

!! Adding
Topics
Deleted lines 22-25:
!Gamma

* write text for home page, howto

Added line 26:
* conferences: IWSLT 2011
Changed line 30 from:
* paper stats, trending topics
to:
* paper stats, trending topics
May 04, 2012, at 02:59 AM by Philipp Koehn -
Added lines 10-18:
!! Improving the Descriptions
Currently, the text for most topics is quite rough, since it was pieced together sentence by sentence whenever a new paper appeared. A better structure would be introducing the main concepts alongside the key papers would be helpful before discussing incremental improvements. However, do not remove papers because you deem them not good enough. The goal is to cite all papers and summarize their key contribution. Please read the [[Meta/style guide]].

!! Adding citations
Citations may be added by pasting their BibTex entries into a [[text box -> Meta/AddBibTeX]]. Please ensure that the publication is not already included under a different BibTex label. If the label is the same, then duplicates are automatically detected. If an online source for the pdf file of the paper is known, include it with BibTex field @@url@@.

!! Topics
The topic structure may be [[edited -> Meta/AddBibTeX]], by adding new topics, re-arranging the hierarchy, but there is currently no support for renaming topics via the Wiki. Some of the larger topics should be broken up.

Deleted line 29:
May 04, 2012, at 02:50 AM by Philipp Koehn -
Deleted line 12:
* integrate reviews
May 03, 2012, at 06:05 AM by Philipp Koehn -
Deleted line 13:
* editing of bibtex entries
May 02, 2012, at 07:26 AM by Philipp Koehn -
Deleted line 22:
* subscribe (to topics), daily email alert with changes
April 28, 2012, at 08:48 PM by Philipp Koehn -
Added line 21:
* cache pdfs locally, allow download individually or all of them (per topic?) in a linked pdf
April 28, 2012, at 08:17 PM by Philipp Koehn -
Deleted lines 11-12:
A lot to do:
* write intros for all topics
October 10, 2011, at 01:25 AM by Philipp Koehn -
Changed line 14 from:
* write text for home page, howto, style guide
to:
* write text for home page, howto
October 04, 2011, at 08:14 PM by Philipp Koehn -
Deleted line 16:
* search for authors, papers in bibtex
October 04, 2011, at 12:53 PM by Philipp Koehn -
Deleted line 15:
* assign most papers to topics
October 03, 2011, at 05:19 PM by Philipp Koehn -
Changed line 7 from:
* adding citations
to:
* [[adding citations -> Meta/AddBibTeX]]
October 03, 2011, at 05:18 PM by Philipp Koehn -
Changed line 8 from:
* [[adding topics -> Meta/Edit Topics]]
to:
* [[adding topics -> Meta/Topics]]
October 03, 2011, at 05:18 PM by Philipp Koehn -
Added lines 1-9:
!Welcome

The goal of this Wiki is to cover all publications on all topics in the field of statistical machine translation.

Please contribute by
* improving the descriptions
* adding citations
* [[adding topics -> Meta/Edit Topics]]

October 03, 2011, at 04:15 PM by Philipp Koehn -
Changed line 5 from:
* write text to home page, howto, style guide
to:
* write text for home page, howto, style guide
October 03, 2011, at 04:15 PM by Philipp Koehn -
Added line 5:
* write text to home page, howto, style guide
October 03, 2011, at 04:13 PM by Philipp Koehn -
Changed line 5 from:
* integrate reviews on paper
to:
* integrate reviews
October 03, 2011, at 04:13 PM by Philipp Koehn -
Added lines 4-6:
* write intros for all topics
* integrate reviews on paper
* assign most papers to topics
October 03, 2011, at 04:12 PM by Philipp Koehn -
Changed lines 1-2 from:
!Beta
to:
!Gamma
Added lines 5-10:
* search for authors, papers in bibtex

!Beta

More to do:
* contact key people for comments
Deleted line 11:
* search for authors, papers
October 03, 2011, at 12:20 AM by Philipp Koehn -
Deleted line 3:
* add new papers
October 02, 2011, at 07:22 PM by Philipp Koehn -
Added line 5:
* editing of bibtex entries
October 02, 2011, at 06:44 PM by Philipp Koehn -
Deleted lines 9-13:
!!Publications
!!Benchmarks
!!Discussion
!!Related Topics
!!New Publications
October 01, 2011, at 11:33 PM by Philipp Koehn -
Deleted line 3:
* {-adding everything from the book-}
Deleted lines 4-5:
* {-categories at bottom-}
* {-automatic detection of uncited papers-}
October 01, 2011, at 11:33 PM by Philipp Koehn -
October 01, 2011, at 11:32 PM by Philipp Koehn -
Changed line 4 from:
* adding everything from the book
to:
* {-adding everything from the book-}
Changed lines 6-7 from:
* categories at bottom
* automatic detection of uncited papers
to:
* {-categories at bottom-}
* {-automatic detection of uncited papers-}
October 01, 2011, at 07:08 PM by Philipp Koehn -
Changed lines 10-11 from:
* Paper stats, trending topics
to:
* paper stats, trending topics
* subscribe (to topics), daily email alert with changes
October 01, 2011, at 04:41 PM by Philipp Koehn -
Added line 10:
* Paper stats, trending topics
October 01, 2011, at 04:36 PM by Philipp Koehn -
Added line 12:
!!Benchmarks
October 01, 2011, at 04:36 PM by Philipp Koehn -
Changed line 12 from:
!!Discussion / Open Questions
to:
!!Discussion
October 01, 2011, at 04:36 PM by Philipp Koehn -
Changed line 14 from:
!!New Papers
to:
!!New Publications
October 01, 2011, at 04:36 PM by Philipp Koehn -
Changed line 11 from:
!!Published Research
to:
!!Publications
October 01, 2011, at 04:35 PM by Philipp Koehn -
Changed lines 11-12 from:

to:
!!Published Research
!!Discussion / Open Questions
!!Related Topics
!!New Papers
October 01, 2011, at 04:35 PM by Philipp Koehn -
Added line 6:
* categories at bottom
Added line 10:
October 01, 2011, at 04:34 PM by Philipp Koehn -
Changed lines 1-2 from:
!Decoding for Phrase-Based Models
to:
!Beta

A lot to do:
* adding everything from the book
* add new papers
Deleted lines 8-10:

!!Stack decoding
The common stack decoding algorithm has it roots in work by (:cite Tillmann:1997 :), who describe a dynamic programming algorithm, similar to techniques for hidden Markov models, for monotone decoding of word-based models. Allowing for reordering makes decoding more complex. Efficient A* search makes the translation of short sentences possible (:citep Och:2001a :). However, longer sentences require pruning (:citep Wang:1997 Niessen:1998 :) or restrictions on reordering (:citep Tillmann:2000 :) or both (:citep Tillmann:2003j :).  (:cite ortizmartinez-garciavarea-casacuberta:2006:WMT :) compare different types of stack decoding algorithms with varying number of stacks. (:cite Delaney:2006:IWSLT :) speed up stack decoding with A* pruning. (:cite Moore:2007:MTSummit :) stress the importance of threshold pruning and the avoidance of expanding doomed hypotheses. Some efficiency may be gained by collapsing contexts that are invariant to the language model (:citep li-khudanpur:2008:SSST :).
October 01, 2011, at 04:33 PM by Philipp Koehn -
Changed lines 1-5 from:
!Stack decoding

Definition

!!Published Research
to:
!Decoding for Phrase-Based Models

* automatic detection of uncited papers
* automatic finding of pdfs for papers
* search for authors, papers

!!Stack decoding
Changed lines 10-16 from:
!!Open Problems

!!New Publications
* (:cite Delaney:2006:IWSLT :)
* (:cite Moore:2007:MTSummit :)

|| Beam sizes ||
to:
October 01, 2011, at 04:38 AM by Philipp Koehn -
Changed line 5 from:
!!Related Work
to:
!!Published Research
Added lines 13-14:

|| Beam sizes ||
October 01, 2011, at 04:23 AM by Philipp Koehn -
Added lines 11-12:
* (:cite Delaney:2006:IWSLT :)
* (:cite Moore:2007:MTSummit :)
October 01, 2011, at 04:21 AM by Philipp Koehn -
Changed lines 1-6 from:
!Decoding for Phrase-Based Models

* automatic detection of uncited papers
* automatic finding of pdfs for papers

!!Stack decoding
to:
!Stack decoding

Definition

!!Related Work
Changed lines 8-10 from:
to:
!!Open Problems

!!New Publications
October 01, 2011, at 01:52 AM by 129.215.26.86 -
Added line 8:
October 01, 2011, at 01:42 AM by 129.215.26.86 -
Changed line 7 from:
The stack decoding algorithm presented here has it roots in work by (:cite Tillmann:1997 :), who describe a dynamic programming algorithm, similar to techniques for hidden Markov models, for monotone decoding of word-based models. Allowing for reordering makes decoding more complex. Efficient A* search makes the translation of short sentences possible (:citep Och:2001a :). However, longer sentences require pruning (:citep Wang:1997 Niessen:1998 :) or restrictions on reordering (:citep Tillmann:2000 :) or both (:citep Tillmann:2003j :).  (:cite ortizmartinez-garciavarea-casacuberta:2006:WMT :) compare different types of stack decoding algorithms with varying number of stacks. (:cite Delaney:2006:IWSLT :) speed up stack decoding with A* pruning. (:cite Moore:2007:MTSummit :) stress the importance of threshold pruning and the avoidance of expanding doomed hypotheses. Some efficiency may be gained by collapsing contexts that are invariant to the language model (:citep li-khudanpur:2008:SSST :).
to:
The common stack decoding algorithm has it roots in work by (:cite Tillmann:1997 :), who describe a dynamic programming algorithm, similar to techniques for hidden Markov models, for monotone decoding of word-based models. Allowing for reordering makes decoding more complex. Efficient A* search makes the translation of short sentences possible (:citep Och:2001a :). However, longer sentences require pruning (:citep Wang:1997 Niessen:1998 :) or restrictions on reordering (:citep Tillmann:2000 :) or both (:citep Tillmann:2003j :).  (:cite ortizmartinez-garciavarea-casacuberta:2006:WMT :) compare different types of stack decoding algorithms with varying number of stacks. (:cite Delaney:2006:IWSLT :) speed up stack decoding with A* pruning. (:cite Moore:2007:MTSummit :) stress the importance of threshold pruning and the avoidance of expanding doomed hypotheses. Some efficiency may be gained by collapsing contexts that are invariant to the language model (:citep li-khudanpur:2008:SSST :).
October 01, 2011, at 01:39 AM by 129.215.26.86 -
Changed line 7 from:
The stack decoding algorithm presented here has it roots in work by (:cite Tillmann:1997 :), who describe a dynamic programming algorithm, similar to techniques for hidden Markov models, for monotone decoding of word-based models. Allowing for reordering makes decoding more complex. Efficient A* search makes the translation of short sentences possible (:citep Och:2001a :). However, longer sentences require pruning (:citep Wang:1997 Niessen:1998 :) or restrictions on reordering (:citep Tillmann:2000 :) or both (:citep Tillmann:2003j :).  (:cite ortizmartinez-garciavarea-casacuberta:2006:WMT :) compare different type of stack decoding with varying number of stacks. (:cite Delaney:2006:IWSLT :) speed up stack decoding with A* pruning. (:cite Moore:2007:MTSummit :) stress the importance of threshold pruning and the avoidance of expanding doomed hypotheses. Some efficiency may be gained by collapsing contexts that are invariant to the language model (:citep li-khudanpur:2008:SSST :).
to:
The stack decoding algorithm presented here has it roots in work by (:cite Tillmann:1997 :), who describe a dynamic programming algorithm, similar to techniques for hidden Markov models, for monotone decoding of word-based models. Allowing for reordering makes decoding more complex. Efficient A* search makes the translation of short sentences possible (:citep Och:2001a :). However, longer sentences require pruning (:citep Wang:1997 Niessen:1998 :) or restrictions on reordering (:citep Tillmann:2000 :) or both (:citep Tillmann:2003j :).  (:cite ortizmartinez-garciavarea-casacuberta:2006:WMT :) compare different types of stack decoding algorithms with varying number of stacks. (:cite Delaney:2006:IWSLT :) speed up stack decoding with A* pruning. (:cite Moore:2007:MTSummit :) stress the importance of threshold pruning and the avoidance of expanding doomed hypotheses. Some efficiency may be gained by collapsing contexts that are invariant to the language model (:citep li-khudanpur:2008:SSST :).
October 01, 2011, at 01:36 AM by 129.215.26.86 -
Changed line 6 from:
!!6.1 Stack decoding
to:
!!Stack decoding
October 01, 2011, at 01:36 AM by 129.215.26.86 -
Changed line 1 from:
!Survey
to:
!Decoding for Phrase-Based Models
October 01, 2011, at 01:35 AM by 129.215.26.86 -
Changed line 7 from:
The stack decoding algorithm presented here has it roots in work by (:cite Tillmann:1997 :), who describe a dynamic programming algorithm, similar to techniques for hidden Markov models, for monotone decoding of word-based models. Allowing for reordering makes decoding more complex. Efficient A* search makes the translation of short sentences possible (:citep Och:2001a :). However, longer sentences require pruning (:citep Wang:1997 Niessen:1998 :) or restrictions on reordering (:citep Tillmann:2000 :) or both (:citep Tillmann:2003j :).  (:cite ortizmartinez-garciavarea-casacuberta:2006:WMT :) compare different type of stack decoding with varying number of stacks. (:cite Delaney:2006:IWSLT :) speed up stack decoding with A* pruning. (:cite Moore:2007:MTSummit :) stress the importance of threshold pruning and the avoidance of expanding doomed hypotheses (:citep Moore:2007:MTSummit :). Some efficiency may be gained by collapsing contexts that are invariant to the language model (:citep li-khudanpur:2008:SSST :).
to:
The stack decoding algorithm presented here has it roots in work by (:cite Tillmann:1997 :), who describe a dynamic programming algorithm, similar to techniques for hidden Markov models, for monotone decoding of word-based models. Allowing for reordering makes decoding more complex. Efficient A* search makes the translation of short sentences possible (:citep Och:2001a :). However, longer sentences require pruning (:citep Wang:1997 Niessen:1998 :) or restrictions on reordering (:citep Tillmann:2000 :) or both (:citep Tillmann:2003j :).  (:cite ortizmartinez-garciavarea-casacuberta:2006:WMT :) compare different type of stack decoding with varying number of stacks. (:cite Delaney:2006:IWSLT :) speed up stack decoding with A* pruning. (:cite Moore:2007:MTSummit :) stress the importance of threshold pruning and the avoidance of expanding doomed hypotheses. Some efficiency may be gained by collapsing contexts that are invariant to the language model (:citep li-khudanpur:2008:SSST :).
Deleted line 2:
* javascript clicks
Deleted line 2:
* latex2html
Deleted lines 2-4:

! TO DO

Deleted lines 6-26:

(:cite BrownEtAl:1990 :)

(:cite Brown:1988 :)

(:cite BLEU :)

(:cite Koehn:2003np :)

(:cite Koehn:2003c :)

(:cite Eisner:2003 :)

(:citep Koehn:2003c :)

(:citep Eisner:2003 :)

(:cite Eisner:2003 BLEU :)

(:citep Eisner:2003 BLEU :)

Added line 9:
* automatic finding of pdfs for papers
Added lines 2-8:


! TO DO

* latex2html
* javascript clicks
* automatic detection of uncited papers
Changed lines 21-26 from:
(:citep Eisner:2003 BLEU :)
to:
(:citep Eisner:2003 BLEU :)


!!6.1 Stack decoding
The stack decoding algorithm presented here has it roots in work by (:cite Tillmann:1997 :), who describe a dynamic programming algorithm, similar to techniques for hidden Markov models, for monotone decoding of word-based models. Allowing for reordering makes decoding more complex. Efficient A* search makes the translation of short sentences possible (:citep Och:2001a :). However, longer sentences require pruning (:citep Wang:1997 Niessen:1998 :) or restrictions on reordering (:citep Tillmann:2000 :) or both (:citep Tillmann:2003j :).  (:cite ortizmartinez-garciavarea-casacuberta:2006:WMT :) compare different type of stack decoding with varying number of stacks. (:cite Delaney:2006:IWSLT :) speed up stack decoding with A* pruning. (:cite Moore:2007:MTSummit :) stress the importance of threshold pruning and the avoidance of expanding doomed hypotheses (:citep Moore:2007:MTSummit :). Some efficiency may be gained by collapsing contexts that are invariant to the language model (:citep li-khudanpur:2008:SSST :).

Changed lines 19-21 from:
(:cite Eisner:2003 BLEU :)
to:
(:cite Eisner:2003 BLEU :)

(:citep
Eisner:2003 BLEU :)
Added lines 18-19:

(:cite Eisner:2003 BLEU :)
Added lines 15-17:
(:citep Koehn:2003c :)

(:citep Eisner:2003 :)
Added lines 10-11:

(:cite Koehn:2003c :)
Added lines 10-11:

(:cite Eisner:2003 :)
Added lines 6-10:

(:cite BLEU :)

(:cite Koehn:2003np :)

Added lines 4-5:

(:cite Brown:1988 :)
Changed lines 1-12 from:
Welcome to PmWiki!

A local copy of PmWiki's
documentation has been installed along with the software,
and is available via the [[PmWiki/documentation index]]. 

To continue setting up PmWiki, see [[PmWiki/initial setup tasks]].

The [[PmWiki/basic editing]] page describes how to create pages
in PmWiki.  You can practice editing in the [[wiki sandbox]].

More information about PmWiki is available from http
://www.pmwiki.org.
to:
!Survey

(:cite BrownEtAl:1990
:)

Actions

Download

Contribute