def _insert_new_terms_to_tables(self):
"""
Insert new terms to tables
clear candidates lists from previous iteration
"""
self.opinion_candidate_list_prev_iter = {}
self.opinion_candidate_list_raw = _merge_tables(
self.opinion_candidate_list_raw, self.opinion_candidate_list_curr_iter
)
for cand_term_list in self.opinion_candidate_list_curr_iter.values():
if len(cand_term_list) >= self.min_freq_opinion_candidate:
new_opinion_term = _set_opinion_term_polarity(cand_term_list)
self.opinion_candidate_list_prev_iter[str(new_opinion_term)] = new_opinion_term
self.opinion_candidate_list_curr_iter = {}
self.opinion_candidate_list = {
**self.opinion_candidate_list,
**self.opinion_candidate_list_prev_iter,
}
self.aspects_candidate_list_prev_iter = list()
self.aspect_candidate_list_raw = _merge_tables(
self.aspect_candidate_list_raw, self.aspect_candidate_list_curr_iter
)
for extracted_aspect_list in self.aspect_candidate_list_curr_iter.values():
if len(extracted_aspect_list) >= self.min_freq_aspect_candidate:
first = extracted_aspect_list[0]
new_aspect_entry = AspectTerm(first.term, first.pos, first.lemma)
if new_aspect_entry not in self.aspects_candidate_list_prev_iter:
self.aspects_candidate_list_prev_iter.append(new_aspect_entry)
self.aspect_candidate_list_curr_iter = {}
self.aspect_candidate_list = (
self.aspect_candidate_list + self.aspects_candidate_list_prev_iter
)
def extract_terms_from_doc(self, parsed_doc):
"""Extract candidate terms for sentences in parsed document.
Args:
parsed_doc (ParsedDocument): Input parsed document.
"""
for text, parsed_sent in parsed_doc.sent_iter():
relations = _get_rel_list(parsed_sent)
for rel_entry in relations:
if rel_entry.rel != 'root':
gov_seen = self.opinion_candidate_list_prev_iter.get(
rel_entry.gov.text)
dep_seen = self.opinion_candidate_list_prev_iter.get(
rel_entry.dep.text)
opinions = []
aspects = []
# =========================== acquisition rules ==============================
if bool(gov_seen) ^ bool(dep_seen):
opinions.append(
rule_1(rel_entry, gov_seen, dep_seen, text))
if not gov_seen and dep_seen:
opinions.append(
rule_2(rel_entry, relations, dep_seen, text))
aspects.append(rule_3(rel_entry, relations, text))
aspects.append(rule_4(rel_entry, relations, text))
if self.aspects_candidate_list_prev_iter and \
AspectTerm.from_token(rel_entry.gov) \
in self.aspects_candidate_list_prev_iter and \
AspectTerm.from_token(rel_entry.dep) \
not in self.aspects_candidate_list_prev_iter:
opinions.append(rule_5(rel_entry, text))
aspects.append(rule_6(rel_entry, relations, text))
self._add_opinion_term(opinions)
self._add_aspect_term(aspects)
def _add_aspect_term(self, terms):
"""
add new aspect term to table.
Args:
terms (list of CandidateTerm): candidate terms list
"""
for term in terms:
if term:
term_entry = AspectTerm(term.term, term.pos)
if term_entry not in self.init_aspect_dict and \
term_entry not in self.aspect_candidate_list and not\
self.ASPECT_STOP_LIST.is_in_stop_list(term.term[0]):
_insert_new_term_to_table(
term, self.aspect_candidate_list_curr_iter)
return True