[jira] [Commented] (JOSHUA-313) Provide a language model based on OpenNLP

2016-10-12 Thread Tommaso Teofili (JIRA)

[ 
https://issues.apache.org/jira/browse/JOSHUA-313?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15568776#comment-15568776
 ] 

Tommaso Teofili commented on JOSHUA-313:


the main point of interest there is we should be able to make improvements 
quite easily there (on Apache OpenNLP project) while I'm not sure of how much 
that'd be possible with BerkeleyLM; anyway I agree that at the current state it 
wouldn't give us performance / accuracy improvements.

> Provide a language model based on OpenNLP
> -
>
> Key: JOSHUA-313
> URL: https://issues.apache.org/jira/browse/JOSHUA-313
> Project: Joshua
>  Issue Type: Improvement
>  Components: core
>Reporter: Tommaso Teofili
>Priority: Minor
> Fix For: 7
>
>
> Since OPENNLP-659 OpenNLP has language modelling capabilities so we could 
> evaluate it within Joshua.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (JOSHUA-314) Enable set structured-output from config file

2016-10-12 Thread Tommaso Teofili (JIRA)

[ 
https://issues.apache.org/jira/browse/JOSHUA-314?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15568733#comment-15568733
 ] 

Tommaso Teofili commented on JOSHUA-314:


+1 sounds good to me :)

> Enable set structured-output from config file
> -
>
> Key: JOSHUA-314
> URL: https://issues.apache.org/jira/browse/JOSHUA-314
> Project: Joshua
>  Issue Type: Improvement
>  Components: core
>Reporter: Tommaso Teofili
>
> Currently if one sets _use-structured-output = true_ in joshua.config that 
> results in error when parsing the config as it's not explicitly handled by 
> {{JoshuaConfiguration#readConfig}} (it can only be set programmatically), I 
> think it'd be nice to be able to configure it from config file too.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (JOSHUA-313) Provide a language model based on OpenNLP

2016-10-12 Thread Matt Post (JIRA)

[ 
https://issues.apache.org/jira/browse/JOSHUA-313?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15568747#comment-15568747
 ] 

Matt Post commented on JOSHUA-313:
--

What specifically did you have in mind? From a quick look it seems that the 
abilities there are fairly limited, and in fact they might benefit from 
importing BerkeleyLM (which is not as good as KenLM for training LMs or 
representing them, but is good, and is Apache-licensed).

> Provide a language model based on OpenNLP
> -
>
> Key: JOSHUA-313
> URL: https://issues.apache.org/jira/browse/JOSHUA-313
> Project: Joshua
>  Issue Type: Improvement
>  Components: core
>Reporter: Tommaso Teofili
>Priority: Minor
> Fix For: 7
>
>
> Since OPENNLP-659 OpenNLP has language modelling capabilities so we could 
> evaluate it within Joshua.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (JOSHUA-314) Enable set structured-output from config file

2016-10-12 Thread Matt Post (JIRA)

[ 
https://issues.apache.org/jira/browse/JOSHUA-314?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15568730#comment-15568730
 ] 

Matt Post commented on JOSHUA-314:
--

It really bothers me (JOSHUA-289 in particular), but I was just going to put it 
off to 6.2 (which will be 7) because it's not really crucial and will just 
release 6.1 further. It's not really much of an advertised feature and we have 
lots of plans toward the API for 7, so it makes sense to me to just ignore it 
for now. Does that sound okay? 


> Enable set structured-output from config file
> -
>
> Key: JOSHUA-314
> URL: https://issues.apache.org/jira/browse/JOSHUA-314
> Project: Joshua
>  Issue Type: Improvement
>  Components: core
>Reporter: Tommaso Teofili
>
> Currently if one sets _use-structured-output = true_ in joshua.config that 
> results in error when parsing the config as it's not explicitly handled by 
> {{JoshuaConfiguration#readConfig}} (it can only be set programmatically), I 
> think it'd be nice to be able to configure it from config file too.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (JOSHUA-314) Enable set structured-output from config file

2016-10-12 Thread Tommaso Teofili (JIRA)

[ 
https://issues.apache.org/jira/browse/JOSHUA-314?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15568719#comment-15568719
 ] 

Tommaso Teofili commented on JOSHUA-314:


thanks a lot Matt for the clarification, then I think this doesn't really make 
much sense, or maybe just for 6.x release?

> Enable set structured-output from config file
> -
>
> Key: JOSHUA-314
> URL: https://issues.apache.org/jira/browse/JOSHUA-314
> Project: Joshua
>  Issue Type: Improvement
>  Components: core
>Reporter: Tommaso Teofili
>
> Currently if one sets _use-structured-output = true_ in joshua.config that 
> results in error when parsing the config as it's not explicitly handled by 
> {{JoshuaConfiguration#readConfig}} (it can only be set programmatically), I 
> think it'd be nice to be able to configure it from config file too.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (JOSHUA-314) Enable set structured-output from config file

2016-10-12 Thread Matt Post (JIRA)

[ 
https://issues.apache.org/jira/browse/JOSHUA-314?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15568694#comment-15568694
 ] 

Matt Post commented on JOSHUA-314:
--

Agreed that this is a problem, but the plan for 7 is to get rid of the 
structured / non-structured distinction entirely, so that structured output is 
*always* what is returned. The output formatting is currently a huge mess, with 
redundant options all over the place, and we are going to clean that up (see 
JOSHUA-289).

https://issues.apache.org/jira/browse/JOSHUA-289


> Enable set structured-output from config file
> -
>
> Key: JOSHUA-314
> URL: https://issues.apache.org/jira/browse/JOSHUA-314
> Project: Joshua
>  Issue Type: Improvement
>  Components: core
>Reporter: Tommaso Teofili
>
> Currently if one sets _use-structured-output = true_ in joshua.config that 
> results in error when parsing the config as it's not explicitly handled by 
> {{JoshuaConfiguration#readConfig}} (it can only be set programmatically), I 
> think it'd be nice to be able to configure it from config file too.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)