Update app.py
Browse files
app.py
CHANGED
|
@@ -26,7 +26,7 @@ def process_data(input_data,columnname = 'text', num_data = 100):
|
|
| 26 |
random.seed(20979738)
|
| 27 |
processed_data = [i for i in input_data[columnname]]
|
| 28 |
random_selection = random.sample(processed_data, num_data)
|
| 29 |
-
filtered_data = filter_similar_items(random_selection, similarity_threshold = 0.
|
| 30 |
st.write('Number of data input: ',len(random_selection))
|
| 31 |
st.write('After removing duplicates: ',len(filtered_data))
|
| 32 |
return filtered_data
|
|
|
|
| 26 |
random.seed(20979738)
|
| 27 |
processed_data = [i for i in input_data[columnname]]
|
| 28 |
random_selection = random.sample(processed_data, num_data)
|
| 29 |
+
filtered_data = filter_similar_items(random_selection, similarity_threshold = 0.75)
|
| 30 |
st.write('Number of data input: ',len(random_selection))
|
| 31 |
st.write('After removing duplicates: ',len(filtered_data))
|
| 32 |
return filtered_data
|