def tweetData(year):
tweets = pull_text.Data(year).tweets
return tweets
def get_nominees(year):
key_words = [
'nominated', 'nominate', 'nominee', 'nominates', 'nominating',
'nominations', 'nomination', 'nom', 'noms', 'nommed'
]
tweets = pull_text.Data(year).tweets
if year == 2013 or year == 2015:
awards = OFFICIAL_AWARDS_1.copy()
else:
awards = OFFICIAL_AWARDS_2.copy()
# key_words.extend(awards)
award_candidate_sents = dict()
for tweet in tweets:
try:
cont_nom = False
for word in key_words:
if fuzz.partial_token_sort_ratio(
word, tweet.text
) > 90: #if the tweet doesn't contain one of the key_words, go to the next tweet
cont_nom = True
break
if not cont_nom:
continue
except:
continue
high_match_idx = -1
high_match_score = -1
equal_scores = []
for idx, award in enumerate(awards):
score = fuzz.token_set_ratio(tweet.text_unchanged, award)
if score > high_match_score:
high_match_idx = idx
high_match_score = score
equal_scores = []
elif score == high_match_score:
equal_scores.append(awards[high_match_idx])
high_match_idx = idx
high_match_score = score
if high_match_score > 50:
if len(equal_scores) > 0:
for candidate in equal_scores:
award_candidate_sents = handle_high_score(
award_candidate_sents, candidate, tweet.text_unchanged,
high_match_score)
award_candidate_sents = handle_high_score(award_candidate_sents,
awards[high_match_idx],
tweet.text_unchanged,
high_match_score)
# print(award_candidate_sents)
final_nominees = dict()
awards_no_candidates = dict()
for award in awards:
bucket = []
award_tokens = word_tokenize(award)
if 'actor' in award_tokens or 'actress' in award_tokens or 'director' in award_tokens:
person = True
else:
person = False
if award in award_candidate_sents:
all_candidates = award_candidate_sents[award]
else:
similar_award_candidates = awards.copy()
similar_award_candidates.remove(award)
most_similar = ''
high_score = 0
for similar in similar_award_candidates:
score = fuzz.ratio(award, similar)
if score > high_score:
high_score = score
most_similar = similar
awards_no_candidates[award] = most_similar
continue
for candidate in all_candidates:
tokenized = word_tokenize(candidate)
nom_idx = -1
for word in key_words:
for idx, token in enumerate(tokenized):
if word == token:
nom_idx = idx
break
if nom_idx != -1:
break
if nom_idx == 0 or nom_idx == -1:
bucket.extend(
make_string_combinations(tokenized, 'right', person))
elif nom_idx == len(tokenized) - 1:
bucket.extend(
make_string_combinations(tokenized, 'left', person))
else:
bucket.extend(
make_string_combinations(tokenized[0:nom_idx], 'left',
person))
bucket.extend(
make_string_combinations(tokenized[nom_idx + 1:], 'right',
person))
final_nominees[award] = bucket
complete_noms = dict()
for award in awards:
if award in awards_no_candidates:
use_award = awards_no_candidates[award]
bucket = final_nominees[use_award]
dist = FreqDist(bucket)
for noms in dist.most_common(10)[6:]:
if noms[0] not in complete_noms[award]:
complete_noms[award] = [noms[0]]
else:
complete_noms[award].append(noms[0])
else:
bucket = final_nominees[award]
dist = FreqDist(bucket)
nom_list = []
for noms in dist.most_common(5):
nom_list.append(noms[0])
complete_noms[award] = nom_list
# print(complete_noms)
return complete_noms
def get_tweets(year):
tweets = pull_text.Data(year).tweets
tweettext = []
for tweet in tweets:
tweettext.append(tweet.text)
return tweettext