[Translate-cvs] SF.net SVN: translate: [6962] src/trunk/translate/search/terminology.py

[<fri...@us...>]

Revision: 6962
          http://translate.svn.sourceforge.net/translate/?rev=6962&view=rev
Author:   friedelwolff
Date:     2008-01-29 08:07:51 -0800 (Tue, 29 Jan 2008)

Log Message:
-----------
Remove unnecessary import and commented code. Update comments.

Modified Paths:
--------------
    src/trunk/translate/search/terminology.py

Modified: src/trunk/translate/search/terminology.py
===================================================================
--- src/trunk/translate/search/terminology.py	2008-01-29 16:05:37 UTC (rev 6961)
+++ src/trunk/translate/search/terminology.py	2008-01-29 16:07:51 UTC (rev 6962)
@@ -20,7 +20,6 @@
 
 """A class that does terminology matching"""
 
-from translate.search import segment
 import re
 
 # We don't want to miss certain forms of words that only change a little
@@ -36,6 +35,8 @@
                   (" ", "-"),               #pre order / pre-order
                  ]
 
+#TODO: compile regexes
+
 class TerminologyComparer:
     def __init__(self, max_len=500):
         self.MAX_LEN = max_len
@@ -43,12 +44,10 @@
     def similarity(self, a, b, stoppercentage=40):
         """returns the match quality of term b in the text a"""
         # We could segment the words, but mostly it will give less ideal 
-        # results, since we'll miss plurals, etc.  We also can't search for
-        # multiword terms, such as "Free Software"
+        # results, since we'll miss plurals, etc. Then we also can't search for
+        # multiword terms, such as "Free Software". Ideally we should use a 
+        # stemmer, like the Porter stemmer.
         
-        #words = segment.words(a)
-        #if b in words:
-
         # So we just see if the word occurs anywhere. This is not perfect since
         # we might get more than we bargained for. The term "form" will be found
         # in the word "format", for example. A word like "at" will trigger too


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.