Update classification.py
Browse files- classification.py +12 -6
classification.py
CHANGED
@@ -17,9 +17,15 @@ def initialize_models():
|
|
17 |
def generate_embeddings(df, model, Column):
|
18 |
embeddings_list = []
|
19 |
for index, row in df.iterrows():
|
20 |
-
if type(row[
|
21 |
print(index)
|
22 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
23 |
embeddings = model.encode(content, convert_to_tensor=True)
|
24 |
embeddings_list.append(embeddings)
|
25 |
else:
|
@@ -39,13 +45,13 @@ def process_categories(categories, model):
|
|
39 |
|
40 |
|
41 |
|
42 |
-
def match_categories(df, category_df):
|
43 |
|
44 |
categories_list, experts_list, topic_list, scores_list = [], [], [], []
|
45 |
for ebd_content in df['Embeddings']:
|
46 |
if isinstance(ebd_content, torch.Tensor):
|
47 |
cos_scores = util.cos_sim(ebd_content, torch.stack(list(category_df['Embeddings']), dim=0))[0]
|
48 |
-
high_score_indices = [i for i, score in enumerate(cos_scores) if score >
|
49 |
|
50 |
# Append the corresponding categories, experts, and topics for each high-scoring index
|
51 |
categories_list.append([category_df.loc[index, 'description'] for index in high_score_indices])
|
@@ -86,7 +92,7 @@ def save_data(df, filename):
|
|
86 |
df.to_excel(new_filename, index=False)
|
87 |
return new_filename
|
88 |
|
89 |
-
def classification(column, file_path, categories):
|
90 |
# Load data
|
91 |
df = load_data(file_path)
|
92 |
|
@@ -100,7 +106,7 @@ def classification(column, file_path, categories):
|
|
100 |
category_df = process_categories(categories, model_ST)
|
101 |
|
102 |
# Match categories
|
103 |
-
df = match_categories(df, category_df)
|
104 |
|
105 |
# Save data
|
106 |
return save_data(df,file_path), df
|
|
|
17 |
def generate_embeddings(df, model, Column):
|
18 |
embeddings_list = []
|
19 |
for index, row in df.iterrows():
|
20 |
+
if type(row[Column]) == str:
|
21 |
print(index)
|
22 |
+
if 'Title' in df.columns:
|
23 |
+
if type(row["Title"]) == str:
|
24 |
+
content = row["Title"] + "\n" + row[Column]
|
25 |
+
else:
|
26 |
+
content = row[Column]
|
27 |
+
else:
|
28 |
+
content = row[Column]
|
29 |
embeddings = model.encode(content, convert_to_tensor=True)
|
30 |
embeddings_list.append(embeddings)
|
31 |
else:
|
|
|
45 |
|
46 |
|
47 |
|
48 |
+
def match_categories(df, category_df, treshold=0.45):
|
49 |
|
50 |
categories_list, experts_list, topic_list, scores_list = [], [], [], []
|
51 |
for ebd_content in df['Embeddings']:
|
52 |
if isinstance(ebd_content, torch.Tensor):
|
53 |
cos_scores = util.cos_sim(ebd_content, torch.stack(list(category_df['Embeddings']), dim=0))[0]
|
54 |
+
high_score_indices = [i for i, score in enumerate(cos_scores) if score > treshold]
|
55 |
|
56 |
# Append the corresponding categories, experts, and topics for each high-scoring index
|
57 |
categories_list.append([category_df.loc[index, 'description'] for index in high_score_indices])
|
|
|
92 |
df.to_excel(new_filename, index=False)
|
93 |
return new_filename
|
94 |
|
95 |
+
def classification(column, file_path, categories, treshold):
|
96 |
# Load data
|
97 |
df = load_data(file_path)
|
98 |
|
|
|
106 |
category_df = process_categories(categories, model_ST)
|
107 |
|
108 |
# Match categories
|
109 |
+
df = match_categories(df, category_df, treshold=treshold)
|
110 |
|
111 |
# Save data
|
112 |
return save_data(df,file_path), df
|