def factory(cls, topic, id, words, bible):
doc = Doc()
doc.topic = topic
doc.id = id
doc.words = {}
for word in words:
# Update the bible.
if not word in bible.words:
bible.words[word] = dotdict()
bible.words[word].index = len(bible.words) - 1
bible.words[word].docs = [doc]
else:
if not doc in bible.words[word].docs:
bible.words[word].docs.append(doc)
# Update this doc.
if word in doc.words:
doc.words[word] += 1
else:
doc.words[word] = 1
# Increment the number of documents processed.
Doc.count += 1
# No cluster yet!
doc.cluster = None
return doc
def factory(cls, topic, id, words, bible):
doc = Doc()
doc.topic = topic
doc.id = id
doc.words = {}
for word in words:
# Update the bible.
if not word in bible.words:
bible.words[word] = dotdict()
bible.words[word].index = len(bible.words) - 1
bible.words[word].docs = [doc]
else:
if not doc in bible.words[word].docs:
bible.words[word].docs.append(doc)
# Update this doc.
if word in doc.words:
doc.words[word] += 1
else:
doc.words[word] = 1
# Increment the number of documents processed.
Doc.count += 1
# No cluster yet!
doc.cluster = None
return doc
def __init__(self, k):
"""
Init.
Key arguments:
k -- final number of clusters.
"""
self.k = k
self.clusters = []
for i in range(k):
self.clusters.append(dotdict())
self.clusters[-1].docs = []
def __init__(self, k):
"""
Init.
Key arguments:
k -- final number of clusters.
"""
self.k = k
self.clusters = []
for i in range(k):
self.clusters.append(dotdict())
self.clusters[-1].docs = []