[umls-similarity] What is the most scalable way to calculate similarity and relateness?

[yonatanbitt...@gmail.com [umls-similarity]]

Hello. I've succeeded to solve my installation problem .
 

 I want to use UMLS-Similarity and relatedness in a research paper I'm writing. 
 

 In order to do it, I want to calculate (1) similarity and (2) relatedness for 
a big dataset (Can get to 1 million and even more calls).
 

 Now I'm able to calculate similarity using the cmd of umls-similarity.pl: 
 

 E:\umls_perl\UMLS-Similarity-1.47\utils>umls-similarity.pl skull hand 
-user=root -password=###
 Default Settings:
   --precision 4
   --database umls
   --hostname localhost
   --socket /tmp/mysql.sock
 

   --measure path
 

 User Settings:
   --username root
   --password XXXXXXX
 

 

 

 UMLS-Interface Configuration Information:
 (Default Information - no config file)
 

   Sources (SAB):
      MSH
   Relations (REL):
      PAR
      CHD
 

   Sources (SABDEF):
      UMLS_ALL
   Relations (RELDEF):
      UMLS_ALL
 

 

 0.1111<>skull(C0037303)<>hand(C0018563)
 

 But the problem is that it takes a lot of time, several seconds (after it 
built an index on the first call)
 

 So my questions are: 
 1. What is the best way to calculate similarity and relatedness?
 2. Is it possible to use it on a big dataset like the one I have?
 3. Does anyone a code example for doing several calculations? Calculate a 
similarity and relatedness for say, an excel document of pairs? (I'm not really 
familiar with perl, what I will learn whatever needed)
 

 Thanks!