def get_emotion_sequence(posts, EMO_RESOURCES):
emo_seq = []
agg_profile = {}
for post in posts:
clean_post = text_utils.clean_text(post)
sentences = sent_tokenize(clean_post)
for sent in sentences:
profile, seq = get_emotion_profile_per_post(sent, EMO_RESOURCES)
if profile != {}:
max_emotion = max(profile.items(),
key=operator.itemgetter(1))[0]
max_emo_val = max(profile.items(),
key=operator.itemgetter(1))[1]
if max_emo_val > 0:
for key in profile:
if profile[key] == max_emo_val:
emo_seq.append(max_emotion)
if key in agg_profile.keys():
agg_profile[key] = agg_profile[key] + profile[key]
else:
agg_profile[key] = profile[key]
return emo_seq, agg_profile
def cleanText(inpath):
output = r'F:\projects-he\nlp-emotion-analysis-Jeloh\nlp-emotion-analysis-core\data\coronavirus_reddit_raw_comments.xls'
csv_xls.csv_to_xlsx_pd(inpath, output)
wb = load_workbook(inpath)
a_sheet
indata = xlrd.open_workbook(inpath)
book = Workbook(encoding='utf-8')
outdata = book.add_sheet('sheet1')
table = indata.sheets()[0]
nrows = table.nrows
ncols = table.ncols
for i in range(1, nrows):
alldata = table.row_values(i)
#result = alldata[1]
clean_text_1 = text_utils.clean_text(alldata[1])
outdata.write(i, 1, alldata[0])
outdata.write(i, 2, clean_text_1)
outdata.save(
r"F:\projects-he\nlp-emotion-analysis-Jeloh\nlp-emotion-analysis-core\data\AfterClean.csv"
)
'NEGATION': NEGATION_MAP,
'INTENSIFIERS': INTENSIFIER_MAP,
'PHYSICAL': PHYSICAL
}
return EMO_RESOURCES
results_df = pd.DataFrame()
df = pd.read_csv(path)
print(df.columns)
print(df['emotion'].unique())
EMO_RESOURCES = load_emotion_dictionaries()
for i, row in df.iterrows():
# if(i>20):continue
row_dict = row.to_dict()
# print()
sentence = row['utterance']
clean_text_1 = text_utils.clean_text(sentence)
emotion_profile, emo_seq = emotion_extractor.get_emotion_profile_per_post(
clean_text_1, EMO_RESOURCES)
print(emotion_profile)
row_dict.update(emotion_profile)
print(row_dict)
results_df = results_df.append(row_dict, ignore_index=True)
results_df.to_csv(
r"E:\Projects\Emotion_detection_gihan\finbert_experiments\evaluations\twitter_emotion_evaluations.csv"
)
with open('models/emotions/intensifier_vocab.pkl', 'rb') as f:
INTENSIFIER_MAP = pickle.load(f)
with open('models/emotions/negation_vocab.pkl', 'rb') as f:
NEGATION_MAP = pickle.load(f)
with open('models/clinical_info/physical.pkl', 'rb') as f:
PHYSICAL = pickle.load(f)
EMO_RESOURCES = {'EMOTIONS': EMOTION_MAP,
'NEGATION': NEGATION_MAP,
'INTENSIFIERS': INTENSIFIER_MAP,
'PHYSICAL': PHYSICAL}
return EMO_RESOURCES
if __name__ == '__main__':
EMO_RESOURCES = load_emotion_dictionaries()
text_1 = 'still i feel very sad about the unexpected incident. hopefully the pain will be less, and i am grateful for..'
clean_text_1 = text_utils.clean_text(text_1)
emotion_profile = emotion_extractor.get_emotion_profile_per_post(clean_text_1, EMO_RESOURCES)
clinical_info = clinical_info_extractor.get_physical_sym_profile(clean_text_1, EMO_RESOURCES)
keyphrases = keyphrase_extractor.analyze_keyphrases(clean_text_1)
print(keyphrases)
print(clinical_info)
print(emotion_profile)