def find_top_word(self, _class, topk=None):
'''Find the keywords of given _class
Args:
_class: the _class the chek word
topk: top k keywords to find, None means all words
Returns:
a list of topk term
'''
if topk:
heap = TopkHeap(topk)
for term in self.ciindex.get_terms():
heap.push((self.estimate_mi(_class, term), term))
return heap.topk()
else:
res = []
for term in self.ciindex.get_terms():
res.append((self.estimate_mi(_class, term), term))
return sorted(res)[::-1]
def find_top_word(self, _class, topk = None):
'''Find the keywords of given _class
Args:
_class: the _class the chek word
topk: top k keywords to find, None means all words
Returns:
a list of topk term
'''
if topk:
heap = TopkHeap(topk)
for term in self.ciindex.get_terms():
heap.push((self.estimate_mi(_class, term), term))
return heap.topk()
else:
res = []
for term in self.ciindex.get_terms():
res.append((self.estimate_mi(_class, term), term))
return sorted(res)[::-1]
def build(self):
'''compute all terms' top pmi elements
All terms computed is from iindex's get_terms method.
'''
terms = self.iindex.get_terms()
for term in terms:
self.term_pmi[term] = TopkHeap(self.top)
for i in range(len(terms) - 1):
for j in range(i + 1, len(terms)):
pmi = self.compute_pmi(terms[i], terms[j])
self.term_pmi[terms[i]].push(PMIElement(terms[j], pmi))
self.term_pmi[terms[j]].push(PMIElement(terms[i], pmi))
def get_class_keywords(self, _class, top_k):
heap = TopkHeap(top_k)
for term in self.term_num_docs:
heap.push((self.term_cdf[term][_class], term))
return [item[1] for item in heap.topk()]
def top_k_appear(self, k):
heap = TopkHeap(k)
for term in self.get_terms():
heap.push((self.get_word_appear(term), term))
return heap.topk()
