def __init__(self, freq_data, selected_doc=None, context_doc_groups=None, total_cnt=None):
self.phrase_cnt = agg_phrase_cnt(freq_data, selected_doc)
self.phrase_cnt_context = {}
self.sum_cnt = sum(self.phrase_cnt.values())
self.sum_cnt_context = {}
for group, docs in context_doc_groups.items():
if len(docs) == 1:
self.phrase_cnt_context[group] = total_cnt
else:
self.phrase_cnt_context[group] = agg_phrase_cnt(freq_data, docs)
self.sum_cnt_context[group] = sum(self.phrase_cnt_context[group].values())
def compute_freq_portion(self):
phrase_freq_portion = agg_phrase_cnt(self.freq_data, self.selected_doc)
# print phrase_freq_portion
freq_sum = sum(phrase_freq_portion.values())
for phrase in phrase_freq_portion:
phrase_freq_portion[phrase] /= float(freq_sum)
return phrase_freq_portion
def __init__(self, freq_data, selected_doc=None, context_doc_groups=None, total_cnt=None, global_scores=None):
self.phrase_cnt = agg_phrase_cnt(freq_data, selected_doc)
self.phrase_df = agg_phrase_df(freq_data, selected_doc)
self.phrase_cnt_context = {}
self.phrase_df_context = {}
self.max_df = max(self.phrase_df.values())
self.max_df_context = {}
self.dc_context = {}
self.self_dc = len(selected_doc)
self.sum_cnt = sum(self.phrase_cnt.values())
self.sum_cnt_context = {}
self.global_scores = global_scores
for group, docs in context_doc_groups.items():
if len(docs) == 1 and docs[0] == '-1':
self.phrase_cnt_context[group] = total_cnt
else:
self.phrase_cnt_context[group] = agg_phrase_cnt(freq_data, docs)
self.phrase_df_context[group] = agg_phrase_df(freq_data, docs)
self.max_df_context[group] = max(self.phrase_df_context[group].values())
self.dc_context[group] = len(docs)
self.sum_cnt_context[group] = sum(self.phrase_cnt_context[group].values())
def __init__(self, freq_data, parsed_file, selected_doc=None): self.phrase_cnt = agg_phrase_cnt(freq_data, selected_doc) self.parsed_file = parsed_file self.selected_doc = selected_doc