You are viewing a plain text version of this content. The canonical link for it is here.
Posted to java-user@lucene.apache.org by "d2clon@desarrolladores.org" <d2...@desarrolladores.org> on 2004/10/05 11:03:58 UTC

hibernate y el algoritmo para asignar scores a los hits

(first excuse me for my english)

hi people..

we are programming a few tests battery to see how lucene creates the scores in a search into a very simple and controllable documents.

but we don't understand why the results looks not ok for us..

i am going to try to explain, the details of the tests described at botton..

if you look the tests you will see:

in test 1)
query: info:house
lucene returns a 50.00 score for this document
id      info
-----   ------
house2	house noise noise noise

and the same score for this document:
id      info
-----   ------
house3	house noise noise noise noise


in test 2)
query: info:house

lucene returns more score for this document:
id      info
-----   ------
house3	noise noiseso noise noiseso noise noiseso house house house house

than for this one:
id      info
-----   ------
house5	noise noiseso noise noiseso noise noiseso house house house house house house



i have seen lucene's algoritm for creating this scores, i don't understand all but i 
think the results of these searches are not ok for me.. 

i need comments, or urls to study and improve my search method


thanks for all
d2clon





---------------------------------------------------------------
TESTS
---------------------------------------------------------------




we always execute this query, using the StandardAnalizer:
query: info:house


 test 1
---------------------------------------

document list:

id	info
----	-------
house0	house noise
house1	house noise noise
house2	house noise noise noise
house3	house noise noise noise noise
house4	house noise noise noise noise noise
nohouse	noise noiseso noise noiseso noise noiseso 


scores:

id       score
----     -------
house0    62.5	
house1    50.0	
house2    50.0	
house3    43.75	
house4    37.5







 test 2:
------------------------------------

document list:

id	info
----	-------
house0	noise noiseso noise noiseso noise noiseso house
house1	noise noiseso noise noiseso noise noiseso house house
house2	noise noiseso noise noiseso noise noiseso house house house
house3	noise noiseso noise noiseso noise noiseso house house house house
house4	noise noiseso noise noiseso noise noiseso house house house house house
house5	noise noiseso noise noiseso noise noiseso house house house house house house
house6	noise noiseso noise noiseso noise noiseso house house house house house house house
house7	noise noiseso noise noiseso noise noiseso house house house house house house house house
house8	noise noiseso noise noiseso noise noiseso house house house house house house house house house
house9	noise noiseso noise noiseso noise noiseso house house house house house house house house house house
nohouse	noise noiseso noise noiseso noise noiseso noise noiseso noise noiseso noise noiseso nohouse

scores:

id       score
----     -------
house9    79.0569
house8    75.0
house7    70.7107
house6    66.1438
house3    62.5
house5    61.2372
house4    55.9017
house2    54.1266
house1    44.1942
house0    37.5


---------------------------------------------------------------------
To unsubscribe, e-mail: lucene-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: lucene-user-help@jakarta.apache.org