Fix Issue #743, Updated word2vec.n_similarities and test_word2vec.testSimilarities methods #883

pranay360 · 2016-09-25T13:07:16Z

Changes requested in PR #882 completed

word2vec.n_similarities now raise a ZeroDivisionError when atleast 1 empty list is passed to it.
In test_word2vec.testSimilarities added testcases involving empty lists to check if ZeroDivisionError is raised.
pep8 followed

Fixes Issue piskvorky#743, n_similarity method now raises ZeroDivisionError if atleast one empty list is passed to it.

Added new test cases in testSimilarities method which makes sure whether ZeroDivisionError is raised if atleast one empty list is passed to word2vec.n_similarities method Related to fix for issue piskvorky#743

tmylk · 2016-09-25T16:36:56Z

CHANGELOG.md

@@ -1,11 +1,10 @@
 Changes
 =======

-0.13.2, 2016-08-19
-* export_phrases in Phrases model changed. Fixed issue #794 and added test cases in test/test_phrases.py(@AadityaJ,
-[#879](https://github.com/RaRe-Technologies/gensim/pull/879))


Please don't delete from changelog of phrases fix.

I didn't delete it, I had a previous build in which that entry was not in the changelog, anyway I added it. Thank You for marking out the changes.

tmylk · 2016-09-25T16:39:06Z

gensim/models/word2vec.py

+                       matutils.unitvec(array(v2).mean(axis=0)))
+        else:
+            raise ZeroDivisionError('Atleast one of the passed list is empty.')
+            return


no need to return.

It is more concise to just have one statement above

if not (len(ws1) > 0 and len(ws2) > 0) : raise ZeroDivisionError('Atleast one of the passed list is empty.')

@tmylk & @pranay360 – Two concerns: (1) 'at least' is two words; (2) shouldn't ZeroDivisionError be reserved for cases where an actual division-by-zero occurs? (Passing in unusable values – like empty lists – would seem to call for a ValueError instead.)

Agreed with @gojomo .

tmylk · 2016-09-26T07:34:44Z

CHANGELOG.md

-* wordtopics has changed to word_topics in ldamallet, and fixed issue #764. (@bhargavvader, [#771](https://github.com/RaRe-Technologies/gensim/pull/771))
+
+* Fixed issue #743 , In word2vec's n_similarity method if atleast one empty list is passed ZeroDivisionError is raised.
+* export_phrases in Phrases model changed. Fixed issue #794 and added test cases in test/test_phrases.py(@AadityaJ, #879) bigram construction can now support multiple bigrams within one sentence


Please add to future release 0.13.3

tmylk · 2016-09-26T11:12:53Z

CHANGELOG.md

-    - bigram construction can now support multiple bigrams within one sentence
-* wordtopics has changed to word_topics in ldamallet, and fixed issue #764. (@bhargavvader, [#771](https://github.com/RaRe-Technologies/gensim/pull/771))
+
+* export_phrases in Phrases model changed. Fixed issue #794 and added test cases in test/test_phrases.py(@AadityaJ, #879) bigram construction can now support multiple bigrams within one sentence


Is this change necessary?

are you talking about fixing the issue?

tmylk · 2016-09-26T11:31:34Z

CHANGELOG.md

-    - bigram construction can now support multiple bigrams within one sentence
-* wordtopics has changed to word_topics in ldamallet, and fixed issue #764. (@bhargavvader, [#771](https://github.com/RaRe-Technologies/gensim/pull/771))
+
+* export_phrases in Phrases model changed. Fixed issue #794 and added test cases in test/test_phrases.py(@AadityaJ, #879) bigram construction can now support multiple bigrams within one sentence


This pr is about #743. There is no need to change other parts of changelog.md

piskvorky · 2016-10-14T04:01:35Z

gensim/models/word2vec.py

        v1 = [self[word] for word in ws1]
        v2 = [self[word] for word in ws2]
-        return dot(matutils.unitvec(array(v1).mean(axis=0)), matutils.unitvec(array(v2).mean(axis=0)))
+        return dot(matutils.unitvec(array(v1).mean(axis=0)),


We don't use vertical indent in gensim -- change to normal hanging indent (see PEP8: all statements on new lines, one level of indent).

piskvorky · 2016-10-14T04:02:39Z

gensim/models/word2vec.py

@@ -1539,9 +1539,13 @@ def n_similarity(self, ws1, ws2):
          True

        """
+        if not(len(ws1) and len(ws2)):
+            raise ZeroDivisionError('Atleast one of the passed list is empty.')


If input is a list, idiomatic Python is: if not ws1 or not ws2:.

piskvorky · 2016-10-14T04:03:49Z

Code style changes requested; please fix.

This has already been merged, so probably in another PR. Cheers.

pranay360 added 8 commits September 25, 2016 18:18

Updated n_similarity method

809f2ba

Fixes Issue piskvorky#743, n_similarity method now raises ZeroDivisionError if atleast one empty list is passed to it.

Updated testSimilarities method

3084f6b

Added new test cases in testSimilarities method which makes sure whether ZeroDivisionError is raised if atleast one empty list is passed to word2vec.n_similarities method Related to fix for issue piskvorky#743

Update CHANGELOG.md

14ce73f

Update CHANGELOG.md

552cd59

Rename .travis.yml to .temp.yml

a826db2

Create .travis.yml

b77ac47

Delete .temp.yml

ec51a48

Update .travis.yml

1288dcf

pranay360 changed the title ~~Fix Issue #743, Updated gensim.word2vec.n_similarities and gensim.test.test_word2vec.testSimilarities methods~~ Fix Issue #743, Updated word2vec.n_similarities and test_word2vec.testSimilarities methods Sep 25, 2016

pranay360 added 3 commits September 25, 2016 19:41

fixed assertRaises statements, to prevent error

8bc778c

Update test_word2vec.py

3398e5a

fixing python 2.6 compatibility issues with assertRaises

eea3590

tmylk suggested changes Sep 25, 2016

View reviewed changes

pranay360 added 4 commits September 25, 2016 23:42

Added Requested Changes, removing return

5235bb6

Update CHANGELOG.md

8f3e51c

Update CHANGELOG.md

306b2ec

Update word2vec.py

ec49724

tmylk reviewed Sep 26, 2016

View reviewed changes

Update CHANGELOG.md

dc4df1c

tmylk reviewed Sep 26, 2016

View reviewed changes

Update CHANGELOG.md

9285ef9

tmylk merged commit 0c5c5ed into piskvorky:develop Sep 26, 2016

This was referenced Oct 4, 2016

Word2vec n_similarity returning numpy matrix instead of float with empty list #743

Closed

n_similarity() in word2vec and doc2vec raises ValueError if an empty list is passed #761

Closed

piskvorky reviewed Oct 14, 2016

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix Issue #743, Updated word2vec.n_similarities and test_word2vec.testSimilarities methods #883

Fix Issue #743, Updated word2vec.n_similarities and test_word2vec.testSimilarities methods #883

pranay360 commented Sep 25, 2016 •

edited

Loading

tmylk Sep 25, 2016

pranay360 Sep 25, 2016

tmylk Sep 25, 2016

gojomo Oct 9, 2016

piskvorky Oct 14, 2016

tmylk Sep 26, 2016

tmylk Sep 26, 2016

pranay360 Sep 26, 2016

tmylk Sep 26, 2016

piskvorky Oct 14, 2016

piskvorky Oct 14, 2016

piskvorky commented Oct 14, 2016

Fix Issue #743, Updated word2vec.n_similarities and test_word2vec.testSimilarities methods #883

Fix Issue #743, Updated word2vec.n_similarities and test_word2vec.testSimilarities methods #883

Conversation

pranay360 commented Sep 25, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

piskvorky commented Oct 14, 2016

pranay360 commented Sep 25, 2016 •

edited

Loading