opennlp-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jim - FooBar();" <jimpil1...@gmail.com>
Subject Re: Merging the output of multiple name finders
Date Wed, 18 Apr 2012 10:12:11 GMT
On 18/04/12 11:06, Jörn Kottmann wrote:
> On 04/18/2012 11:41 AM, Jim - FooBar(); wrote:
>> I keep the longest span whenever i'm encountering 2 
>> overlapping/intersecting spans of the same type...it makes sense to 
>> do that cos one of them must be wrong...what about different types 
>> though?
>
> When they are intersecting the following span could be longer, thats 
> right. To keep things simple
> I suggested to always keep the first span.
>
> Jörn

well that sounds a bit arbitrary don't you think? It sounds like 
flipping a coin!
also I'm not talking about span length here...if the types are different 
it makes no sense to compare lengths...the chances of any of them being 
the correct one are equal...if we always keep the earliest one it is 
easy to implement but we're hard-coding something with no good 
justification...see what i mean?

Jim

Mime
View raw message