uima-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Marshall Schor <...@schor.com>
Subject Re: KeyPhrase Extraction
Date Fri, 11 Jun 2010 11:10:02 GMT


On 6/11/2010 3:23 AM, Jürgen Jakobitsch wrote:
> hi marshall,
>
> thanks for your reply, i'm actually looking for alternatives
> to kea. not that kea wouldn't work, but i don't like relying
> on one single extractor.
>
> but if i understand your message right, uima itself can't be
> used as a keyphrase extractor (i was thinking so because of
> one of the demos) and the purpose of uima itself is slightly
> different, am i getting you right?
>   

Yes.  UIMA is a component framework, intended to promote putting
independently developed annotators together, with support for scaleout.
-Marshall
> wkr www.turnguard.com/turnguard
>
>
>
> ----- Original Message -----
> From: "Marshall Schor" <msa@schor.com>
> To: user@uima.apache.org
> Sent: Thursday, June 10, 2010 11:14:08 PM
> Subject: Re: KeyPhrase Extraction
>
> Hi,
>
> The keyphrase extraction algorithm (http://www.nzdl.org/Kea/) is open
> source, and probably could be easily "wrapped" to become a UIMA
> annotator. This would let it be used inside UIMA pipelines, and scaled,
> etc., using the UIMA framework.
>
> UIMA would be beneficial if the goal was to combine this analytic with
> other analytics, or to enable scaling this up.
>
> -Marshall
>
> On 6/10/2010 4:17 PM, Jürgen Jakobitsch wrote:
>   
>> hi,
>>
>> i'm completely new to uima and am wondering, if uima
>> if suitable in any way to perform keyphrase extraction
>> from plain text.
>>
>> by keyphrase extraction i mean keyphrases that best represent
>> the text like kea does.
>>
>> are there any experiences using uima for keyphrase extraction?
>>
>> any pointer really appreciated.
>>
>> wkr www.turnguard.com/turnguard
>>
>>
>>     
>   

Mime
View raw message