Upload sus_fls.py
Browse files- sus_fls.py +52 -0
sus_fls.py
ADDED
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from transformers import RobertaTokenizer,pipeline
|
2 |
+
import torch
|
3 |
+
import nltk
|
4 |
+
from nltk.tokenize import sent_tokenize
|
5 |
+
from fin_readability_sustainability import BERTClass, do_predict
|
6 |
+
import pandas as pd
|
7 |
+
import en_core_web_sm
|
8 |
+
|
9 |
+
nltk.download('punkt')
|
10 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
11 |
+
|
12 |
+
#SUSTAINABILITY STARTS
|
13 |
+
tokenizer_sus = RobertaTokenizer.from_pretrained('roberta-base')
|
14 |
+
model_sustain = BERTClass(2, "sustanability")
|
15 |
+
model_sustain.to(device)
|
16 |
+
model_sustain.load_state_dict(torch.load('sustainability_model.bin', map_location=device)['model_state_dict'])
|
17 |
+
|
18 |
+
def get_sustainability(text):
|
19 |
+
df = pd.DataFrame({'sentence':sent_tokenize(text)})
|
20 |
+
actual_predictions_sustainability = do_predict(model_sustain, tokenizer_sus, df)
|
21 |
+
highlight = []
|
22 |
+
for sent, prob in zip(df['sentence'].values, actual_predictions_sustainability[1]):
|
23 |
+
if prob>=4.384316:
|
24 |
+
highlight.append((sent, 'non-sustainable'))
|
25 |
+
elif prob<=1.423736:
|
26 |
+
highlight.append((sent, 'sustainable'))
|
27 |
+
else:
|
28 |
+
highlight.append((sent, '-'))
|
29 |
+
return highlight
|
30 |
+
#SUSTAINABILITY ENDS
|
31 |
+
|
32 |
+
|
33 |
+
##Forward Looking Statement
|
34 |
+
nlp = en_core_web_sm.load()
|
35 |
+
def split_in_sentences(text):
|
36 |
+
doc = nlp(text)
|
37 |
+
return [str(sent).strip() for sent in doc.sents]
|
38 |
+
def make_spans(text,results):
|
39 |
+
results_list = []
|
40 |
+
for i in range(len(results)):
|
41 |
+
results_list.append(results[i]['label'])
|
42 |
+
facts_spans = []
|
43 |
+
facts_spans = list(zip(split_in_sentences(text),results_list))
|
44 |
+
return facts_spans
|
45 |
+
|
46 |
+
fls_model = pipeline("text-classification", model="yiyanghkust/finbert-fls", tokenizer="yiyanghkust/finbert-fls")
|
47 |
+
def fls(text):
|
48 |
+
results = fls_model(split_in_sentences(text))
|
49 |
+
return make_spans(text,results)
|
50 |
+
|
51 |
+
|
52 |
+
|