Commit
•
d76b5ba
1
Parent(s):
fb7f2e1
Move from dev environment
Browse files- functions.py +2 -0
functions.py
CHANGED
|
@@ -60,7 +60,9 @@ def index_documents(df):
|
|
| 60 |
if i % 5000 == 0:
|
| 61 |
print(f'Indexed {i} documents', end='\r')
|
| 62 |
df['title_abs'] = df['title'] + ' ' + df['abstract']
|
|
|
|
| 63 |
all_text = ' '.join(df['title_abs'])
|
|
|
|
| 64 |
term_frequencies = Counter(analyze(all_text))
|
| 65 |
return(ind, doc, term_frequencies)
|
| 66 |
|
|
|
|
| 60 |
if i % 5000 == 0:
|
| 61 |
print(f'Indexed {i} documents', end='\r')
|
| 62 |
df['title_abs'] = df['title'] + ' ' + df['abstract']
|
| 63 |
+
print('Before all text')
|
| 64 |
all_text = ' '.join(df['title_abs'])
|
| 65 |
+
print('After all text')
|
| 66 |
term_frequencies = Counter(analyze(all_text))
|
| 67 |
return(ind, doc, term_frequencies)
|
| 68 |
|