Thanks for pointing this out Ali,

That's a kind of problem I would have faced soon.

Regards,
Helen

--- In [email protected], ali hasnain <ali.hasnain@...> wrote:
>
> Hello Ted,
> 
> Agreed with the value of -1 for the unknown concepts.
> Finger crossed to see the new code with all these fixes.
> 
> Best Regards,
> Ali Hasnain
> 
> 
> 
> 
> 
> 
> 
> On Tue, Apr 17, 2012 at 4:19 PM, ted_pedersen <duluthted@...> wrote:
> 
> > **
> >
> >
> > Hi Ali,
> >
> > We have a fix in the works that will be released later today, stay tuned!
> >
> > However, we will return -1 in the case where one or both of the terms/CUIs
> > is unknown. This is to be consistent with umls-similarity.pl (the command
> > line interface that runs on installed systems) and also to avoid an
> > ambiguity in the value of 0.
> >
> > 0 means that the two terms / CUIs are known to UMLS, but that there is no
> > similarity between them.
> >
> > -1 means that one or both of the terms / CUIs are unknown to the UMLS, and
> > so the measurment isn't really valid.
> >
> > So, we'd like to keep a distinction between these two cases because they
> > are quite distinct. I hope this makes sense.
> >
> > More soon,
> > Ted
> >
> >
> > --- In [email protected], "ShopingPrincess99785"
> > <ali.hasnain@> wrote:
> > >
> > > Hello Ted,
> > >
> > > Thank you very much for your timely response.
> > > I really appreciate.
> > >
> > > It would be good if the similarity/ relatedness score would be
> > calculated as "0" when one or both the concepts are unknown to UMLS.
> > > This can be a simple way of saying that there is no similarity-
> > irrespective of the fact weather the term is unknown or known to UMLS.
> > >
> > > Anyway looking forward for your timely response with the new code that
> > resolves this problem.
> > >
> > > Best Regards,
> > > Ali Hasnain
> > >
> > >
> > >
> > > --- In [email protected], "ted_pedersen" <duluthted@>
> > wrote:
> > > >
> > > > Hi Ali,
> > > >
> > > > Thanks for reporting this, and I can confirm your observation, the
> > program does indeed hang when one of the input terms is unknown to UMLS. We
> > will take a look at this and report back via the mailing list.
> > > >
> > > > Cordially,
> > > > Ted
> > > >
> > > > --- In [email protected], "ShopingPrincess99785"
> > <ali.hasnain@> wrote:
> > > > >
> > > > > Hello,
> > > > >
> > > > > I am using the UMLS Perl implementation from the command line to
> > find the relatedness of two concepts.
> > > > > So it is noticed that it gives some score when I try to find the
> > relatedness for two concepts that are defined in UMLS example is as follow:
> > > > >
> > > > > When I give the command 
> > > > > "query-umls-similarity-webinterface.pl--measure vector drug molecule" 
> > > > > it gives the score value: 0.4835 .
> > > > > Similarly it computes the values of all other concepts available in
> > UMLS but the problem arrives when one of the two concept is asked that is
> > not defined in UMLS, rather it should give value of 0 meaning there is no
> > relatedness the command line become hanged and stops and waited for
> > unlimited time giving no value or even no error.
> > > > >
> > > > > Example is When I give the command "
> > query-umls-similarity-webinterface.pl --measure vector drug ABCD", so the
> > concept ABCD is nothing defined in UMLS, it should give me some error or
> > some numeric value e.g "0" showing no relatedness.
> > > > >
> > > > > Actually I am much relying on the numeric value that comes as an
> > output for the relatedness and what I want is that it should give some
> > value irrespective of what word has been asked. If that word is not
> > available in the UMLS it should give me "0" or at least it should stop and
> > give some message that this word is not available and computes the values
> > for next pair when I give the words in a LOOP.
> > > > >
> > > > >
> > > > > As I am new to PERL, can you please guide me where should I make
> > changes in the PERL code file of ""query-umls-similarity-webinterface.pl""
> > (that comes with the package) so it gives the value of "0" rather than the
> > program hangs and go further.
> > > > >
> > > > >
> > > > > I consider this as a limitation or an error so please send me as an
> > attachment the edited version of the code file of ""
> > query-umls-similarity-webinterface.pl"" or any other that I can replaced
> > with the existing files so I can continue computing relatedness
> > irrespective of the fact weather any of the concept available in the UMLS
> > or not.
> > > > >
> > > > > Looking forward for your timely response.
> > > > >
> > > > > Best Regards,
> > > > > Ali Hasnain
> > > > >
> > > >
> > >
> >
> >  
> >
>


Reply via email to