geof379 / ws4j

Automatically exported from code.google.com/p/ws4j
0 stars 0 forks source link

Java calculateRelatedness() score differs from ws4j App Demo Score #10

Open GoogleCodeExporter opened 8 years ago

GoogleCodeExporter commented 8 years ago
What steps will reproduce the problem?
1. Input "fire", "terminate" literally (not the #sense)
2. Call the calculateRelatedness() in Java
3. The score returned is .28 but the score returned in the app demo is 1.0

What is the expected output? What do you see instead?
I would expect a 1 to 1

What version of the product are you using? On what operating system?
1.0.1

Does the Web version iterate through each of the senses to come up with a more 
concise number OR is it a more updated corpus? ... If it is the later, would 
you be able to make a REST method that returns a JSON object with the RP, LESK, 
etc values instead of HTML?

Original issue reported on code.google.com by matth...@g.ucla.edu on 25 Jun 2014 at 11:04

GoogleCodeExporter commented 8 years ago
After further exploration of the code and the WS4JConfiguration file, I solved 
the problem. Setting the use most frequent Senses to false "setMFS(false)" 
fixes the problem. 

Considering that there's no documentation, I'd personally recommend that this 
feature be set to false by default. I know it slows down the calculation, 
however since there's no documentation, I (personally as a user) would rather 
be figuring out out how to optimize than figure out why a calculation is not 
accurate.

Original comment by jrie...@gmail.com on 26 Jun 2014 at 7:04

GoogleCodeExporter commented 8 years ago
I suffer from the same problem, but can I know how to measure the relatedness 
between two senses (Word1#n#2,Word2#n#5) as the demo do?

Original comment by saad...@gmail.com on 27 Jun 2014 at 9:26

GoogleCodeExporter commented 8 years ago
[deleted comment]
GoogleCodeExporter commented 8 years ago
how to calculate the relatedness between two senses 
(Word1#n#1,Word2#v#2)???same question here!!!!!i tried to work with concepts 
but there's something wrong with calculating LESK measure .... it's always 0. i 
followed this 
link:http://stackoverflow.com/questions/17166298/docs-for-java-ws4j-library

Original comment by arab.mey...@gmail.com on 3 Mar 2015 at 3:33