def test_euclideanDistanceLengthNormalisation(self):
vector1 = ('cricket','is','a','game','played','by','fools','watched','others')
vector1 = feeds.create_word_vector(vector1)
#test data same article, doubled length to test normalisation
vector2 = ('cricket','is','a','game','played','by','fools','watched','others','cricket','is','a','game','played','by','fools','watched','others')
vector2 = feeds.create_word_vector(vector2)
result = processor.get_euclidean_dif(vector1, vector2)
self.assertEqual(result, 0.0)
def test_euclideanDistanceDifferentArticle(self):
vector1 = ('cricket','is','a','game','played','by','fools','watched','others')
vector1 = feeds.create_word_vector(vector1)
#test data same article, doubled length to test normalisation
vector2 = ('bing','bong','boop','billa','zac','time','test','shane','red')
vector2 = feeds.create_word_vector(vector2)aquire
result = processor.get_euclidean_dif(vector1, vector2)
self.assertAlmostEqual(result, 1.0, 4 )
def addToStory(self, item, conn):
if len(self.storyList) == 0:
newStory = Story(item, conn)
self.storyList.append(newStory)
#print 'empty feedList'
return newStory
else:
bestDif = (1, None)
for thing in self.storyList:
dif = get_euclidean_dif(thing.items[0].article.vector, item.article.vector)
if dif < 0.6:
if dif < bestDif[0]:
bestDif = (dif, thing)
#print dif
thing.addArticle(item.article, conn)
return bestDif[1]
#print item
anewStory = Story(item, conn)
self.storyList.append(anewStory)
#print 'no similar stories'
#print 'story: ' + str(anewStory)
return anewStory
def test_euclideanDistanceSameArticleisZero(self):
vector1 = ('cricket','is','a','game','played','by','fools','watched','others')
vector1 = feeds.create_word_vector(vector1)
result = processor.get_euclidean_dif(vector1, vector1)
self.assertEqual(result, 0.0)