You are viewing a plain text version of this content. The canonical link for it is here.

Posted to users@opennlp.apache.org by David Young <dy...@gmail.com> on 2012/08/23 19:25:03 UTC

Getting started with MaxEnt

Hi all I have just started using OpenNLP its great, used the tutorial to
get POS up and running in C# in a day and its very powerful, but there is
no detailed information in the manual on Maxent.

First I tried to load everything as Java src into NetBeans and I ran into
too many errors that I am not familiar with while building and the examples
folder was missed out completely. How do others build from src and which
IDE is the easiest to set-up?

I found a simple tutorial on Maxent using ShartEntropy at CodeProject which
gives this example of training data:
" Day 6 Cold Dry Early Umbrella

    Day 7        Cold        Rainy      Early        Umbrella
    Day 8        Cold        Dry        Late         No_Umbrella
    Day 9        Warm        Rainy      Late         No_Umbrella

Day 10 Warm Dry Late No_Umbrella"

What I want to do is something more like POS but different enough that I
can not train the POSTagger.
Where can I find an example of the data POS feeds to MaxEnt?

Are there interactions such as secondPreviousWord_previousWord_currentWord?

Thanks for the support.

Re: Getting started with MaxEnt

Posted by Jörn Kottmann <ko...@gmail.com>.

Yes, there are just examples. We have POS training
code which reads in a data set, and then transforms it
into an event stream which used for the training.

The events are not written to disk, all this is done via the API.

Jörn

On 08/24/2012 04:23 PM, David Young wrote:
> Thanks I have seen a few tutorials like this. It shows the theory of what
> MaxEnt does behind the scenes and it shows how to use the java code to
> compile the model from the data. But I can not find an example method of
> creating the training data other than the most basic examples of weather
> and sports. For something like POS I wondered how this is formatted. What
> interactions are in the data and how they are listed and so on.
>

Re: Getting started with MaxEnt

Posted by David Young <dy...@gmail.com>.

Thanks I have seen a few tutorials like this. It shows the theory of what
MaxEnt does behind the scenes and it shows how to use the java code to
compile the model from the data. But I can not find an example method of
creating the training data other than the most basic examples of weather
and sports. For something like POS I wondered how this is formatted. What
interactions are in the data and how they are listed and so on.

Re: Getting started with MaxEnt

Posted by Koji Sekiguchi <ko...@r.email.ne.jp>.

Or see my SlideShare post at P.21-22 in:

http://www.slideshare.net/KojiSekiguchi/opennlp-mem-and-perceptron

Note that OPENNLP-516.patch is no longer needed on trunk as it has been committed.

koji
-- 
http://soleami.com/blog/starting-lab-work.html

(12/08/24 2:45), William Colen wrote:
> Apache OpenNLP and the Maxent package use Maven as build and dependency
> management system. You can find some info here:
> http://opennlp.apache.org/building.html
>
> You can search the web to learn how to import a Maven project into
> NetBeans. It is simple, but I can't guide you because I don't use NetBeans.
>
> William
>
> On Thu, Aug 23, 2012 at 2:25 PM, David Young <dy...@gmail.com> wrote:
>
>> Hi all I have just started using OpenNLP its great, used the tutorial to
>> get POS up and running in C# in a day and its very powerful, but there is
>> no detailed information in the manual on Maxent.
>>
>> First I tried to load everything as Java src into NetBeans and I ran into
>> too many errors that I am not familiar with while building and the examples
>> folder was missed out completely. How do others build from src and which
>> IDE is the easiest to set-up?
>>
>> I found a simple tutorial on Maxent using ShartEntropy at CodeProject which
>> gives this example of training data:
>> " Day 6 Cold Dry Early Umbrella
>>
>>      Day 7        Cold        Rainy      Early        Umbrella
>>      Day 8        Cold        Dry        Late         No_Umbrella
>>      Day 9        Warm        Rainy      Late         No_Umbrella
>>
>> Day 10 Warm Dry Late No_Umbrella"
>>
>> What I want to do is something more like POS but different enough that I
>> can not train the POSTagger.
>> Where can I find an example of the data POS feeds to MaxEnt?
>>
>> Are there interactions such as secondPreviousWord_previousWord_currentWord?
>>
>> Thanks for the support.
>>
>

Re: Getting started with MaxEnt

Posted by William Colen <wi...@gmail.com>.

Apache OpenNLP and the Maxent package use Maven as build and dependency
management system. You can find some info here:
http://opennlp.apache.org/building.html

You can search the web to learn how to import a Maven project into
NetBeans. It is simple, but I can't guide you because I don't use NetBeans.

William

On Thu, Aug 23, 2012 at 2:25 PM, David Young <dy...@gmail.com> wrote:

> Hi all I have just started using OpenNLP its great, used the tutorial to
> get POS up and running in C# in a day and its very powerful, but there is
> no detailed information in the manual on Maxent.
>
> First I tried to load everything as Java src into NetBeans and I ran into
> too many errors that I am not familiar with while building and the examples
> folder was missed out completely. How do others build from src and which
> IDE is the easiest to set-up?
>
> I found a simple tutorial on Maxent using ShartEntropy at CodeProject which
> gives this example of training data:
> " Day 6 Cold Dry Early Umbrella
>
>     Day 7        Cold        Rainy      Early        Umbrella
>     Day 8        Cold        Dry        Late         No_Umbrella
>     Day 9        Warm        Rainy      Late         No_Umbrella
>
> Day 10 Warm Dry Late No_Umbrella"
>
> What I want to do is something more like POS but different enough that I
> can not train the POSTagger.
> Where can I find an example of the data POS feeds to MaxEnt?
>
> Are there interactions such as secondPreviousWord_previousWord_currentWord?
>
> Thanks for the support.
>