def convertSingle(self, user_id):
friends = loadFollowingFromDBofUser(user_id)
def _dummy(friend): # insert feature weight
return (friend, 1)
friends = map(_dummy, friends)
return friends
def genCoeffTable(self, effectives, instances):
'''
@param effectives 2d matrix about effective feature (follower ids)
first dim is label, and second is follower id.
@param instances A list of elements where each element is (user_id,
label)
'''
# we build a table to store the coefficients
table = CoeffTable(effectives)
# we do the following work:
# load following list for the instance
# load text for the instance
# parse the text and make into words array
for user_id, label in instances:
# whenever this loop ends, we process a single user.
text, length = self._readTextCrawled(user_id)
converter = TextFeatureConverter()
if text is None:
continue
words_arr = converter.textToSparseTokens(text, False)
#words_arr = converter.normarlizeWordsArr(words_arr,
# length)
following = loadFollowingFromDBofUser(user_id)
following = map(str, following)
for w, value in words_arr:
table.record(label, following, w, value)
table.addTweetLength(label, following, length)
# normalize the word counts.
table.normalize()
return table