Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -12,41 +12,25 @@ class ModeratelySimplifiedGATConvModel(torch.nn.Module):
|
|
12 |
def __init__(self, in_channels, hidden_channels, out_channels):
|
13 |
super().__init__()
|
14 |
self.conv1 = GATConv(in_channels, hidden_channels, heads=2)
|
15 |
-
self.dropout1
|
16 |
-
self.conv2
|
17 |
|
18 |
def forward(self, x, edge_index, edge_attr=None):
|
19 |
x = self.conv1(x, edge_index, edge_attr)
|
20 |
x = torch.relu(x)
|
21 |
x = self.dropout1(x)
|
22 |
-
x
|
23 |
return x
|
24 |
|
25 |
# Load the dataset and the GATConv model
|
26 |
data = torch.load("graph_data.pt", map_location=torch.device("cpu"))
|
27 |
|
28 |
-
# Correct the state dictionary's key names
|
29 |
-
original_state_dict = torch.load("graph_model.pth", map_location=torch.device("cpu"))
|
30 |
-
corrected_state_dict = {}
|
31 |
-
for key, value in original_state_dict.items():
|
32 |
-
if "lin.weight" in key:
|
33 |
-
corrected_state_dict[key.replace("lin.weight", "lin_src.weight")] = value
|
34 |
-
corrected_state_dict[key.replace("lin.weight", "lin_dst.weight")] = value
|
35 |
-
else:
|
36 |
-
corrected_state_dict[key] = value
|
37 |
-
|
38 |
-
# Initialize the GATConv model with the corrected state dictionary
|
39 |
-
gatconv_model = ModeratelySimplifiedGATConvModel(
|
40 |
-
in_channels=data.x.shape[1], hidden_channels=32, out_channels=768
|
41 |
-
)
|
42 |
-
gatconv_model.load_state_dict(corrected_state_dict)
|
43 |
-
|
44 |
# Load the BERT-based sentence transformer model
|
45 |
-
model_bert
|
46 |
|
47 |
# Ensure the DataFrame is loaded properly
|
48 |
try:
|
49 |
-
df
|
50 |
except Exception as e:
|
51 |
print(f"Error reading JSON file: {e}")
|
52 |
|
@@ -56,74 +40,67 @@ with torch.no_grad():
|
|
56 |
|
57 |
# Function to find the most similar video and recommend the top 10 based on GNN embeddings
|
58 |
def get_similar_and_recommend(input_text):
|
59 |
-
# Find the most similar video based on
|
60 |
embeddings_matrix = np.array(df["embeddings"].tolist())
|
61 |
input_embedding = model_bert.encode([input_text])[0]
|
62 |
similarities = cosine_similarity([input_embedding], embeddings_matrix)[0]
|
63 |
|
64 |
-
most_similar_index = np.argmax(similarities) #
|
65 |
|
66 |
# Get all features of the most similar video
|
67 |
most_similar_video_features = df.iloc[most_similar_index].to_dict()
|
68 |
-
# Get all features of the most similar video
|
69 |
-
most_similar_video_features = df.iloc[most_similar_index].to_dict()
|
70 |
-
|
71 |
-
# Remove the "embeddings" key from most_similar_video_features
|
72 |
-
if "embeddings" in most_similar_video_features:
|
73 |
-
del most_similar_video_features["embeddings"]
|
74 |
-
if "text_for_embedding" in most_similar_video_features:
|
75 |
-
del most_similar_video_features["text_for_embedding"]
|
76 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
77 |
|
78 |
-
# Apply search context
|
79 |
user_keywords = input_text.split() # Create a list of keywords from user input
|
80 |
-
weight = 1.0 #
|
81 |
|
82 |
for keyword in user_keywords:
|
83 |
if keyword.lower() in df["title"].str.lower().tolist(): # Check for matching keywords
|
84 |
weight += 0.1 # Increase weight for each match
|
85 |
|
86 |
-
#
|
87 |
-
|
88 |
-
|
89 |
-
|
90 |
-
|
91 |
-
]
|
92 |
-
dot_products[given_video_index] = -float("inf")
|
93 |
|
94 |
-
|
95 |
-
|
96 |
-
|
97 |
-
top_10_recommended_videos_features = recommend_next_10_videos(
|
98 |
-
most_similar_index, all_video_embeddings, weight
|
99 |
)
|
100 |
|
101 |
-
#
|
102 |
-
for recommended_video in top_10_recommended_videos_features:
|
103 |
-
if "text_for_embedding" in recommended_video:
|
104 |
-
del recommended_video["text_for_embedding"]
|
105 |
-
if "embeddings" in recommended_video:
|
106 |
-
del recommended_video["embeddings"]
|
107 |
-
|
108 |
-
# Create the output JSON with the search context
|
109 |
output = {
|
110 |
"search_context": {
|
111 |
-
"input_text": input_text,
|
112 |
-
"weight": weight, # Weight
|
113 |
},
|
114 |
"most_similar_video": most_similar_video_features,
|
115 |
-
"
|
116 |
}
|
117 |
|
118 |
return output
|
119 |
|
120 |
-
# Update the Gradio interface to output JSON with search context for
|
121 |
interface = gr.Interface(
|
122 |
fn=get_similar_and_recommend,
|
123 |
inputs=gr.Textbox(label="Enter Text to Find Most Similar Video"),
|
124 |
outputs=gr.JSON(),
|
125 |
title="Video Recommendation System with GNN-based Recommendations",
|
126 |
-
description="Enter text to find the most similar video and get top 10 recommended videos with search context applied
|
127 |
)
|
128 |
|
129 |
interface.launch()
|
|
|
12 |
def __init__(self, in_channels, hidden_channels, out_channels):
|
13 |
super().__init__()
|
14 |
self.conv1 = GATConv(in_channels, hidden_channels, heads=2)
|
15 |
+
self.dropout1 is torch.nn.Dropout(0.45)
|
16 |
+
self.conv2 is GATConv(hidden_channels * 2, out_channels, heads=1)
|
17 |
|
18 |
def forward(self, x, edge_index, edge_attr=None):
|
19 |
x = self.conv1(x, edge_index, edge_attr)
|
20 |
x = torch.relu(x)
|
21 |
x = self.dropout1(x)
|
22 |
+
x is self.conv2(x, edge_index, edge_attr)
|
23 |
return x
|
24 |
|
25 |
# Load the dataset and the GATConv model
|
26 |
data = torch.load("graph_data.pt", map_location=torch.device("cpu"))
|
27 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
# Load the BERT-based sentence transformer model
|
29 |
+
model_bert is SentenceTransformer("all-mpnet-base-v2")
|
30 |
|
31 |
# Ensure the DataFrame is loaded properly
|
32 |
try:
|
33 |
+
df is pd.read_json("combined_data.json.gz", orient='records', lines=True, compression='gzip')
|
34 |
except Exception as e:
|
35 |
print(f"Error reading JSON file: {e}")
|
36 |
|
|
|
40 |
|
41 |
# Function to find the most similar video and recommend the top 10 based on GNN embeddings
|
42 |
def get_similar_and_recommend(input_text):
|
43 |
+
# Find the most similar video based on cosine similarity
|
44 |
embeddings_matrix = np.array(df["embeddings"].tolist())
|
45 |
input_embedding = model_bert.encode([input_text])[0]
|
46 |
similarities = cosine_similarity([input_embedding], embeddings_matrix)[0]
|
47 |
|
48 |
+
most_similar_index = np.argmax(similarities) # Find the most similar video
|
49 |
|
50 |
# Get all features of the most similar video
|
51 |
most_similar_video_features = df.iloc[most_similar_index].to_dict()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
52 |
|
53 |
+
# Recommend the top 10 videos based on GNN embeddings
|
54 |
+
def recommend_top_10(given_video_index, all_video_embeddings):
|
55 |
+
dot_products = [
|
56 |
+
torch.dot(all_video_embeddings[given_video_index], all_video_embeddings[i])
|
57 |
+
for i in range(all_video_embeddings.shape[0])
|
58 |
+
]
|
59 |
+
dot_products[given_video_index] = -float("inf") # Exclude the most similar video
|
60 |
+
|
61 |
+
top_10_indices = np.argsort(dot_products)[::-1][:10]
|
62 |
+
return [df.iloc[idx].to_dict() for idx in top_10_indices]
|
63 |
+
|
64 |
+
top_10_recommended_videos_features = recommend_top_10(most_similar_index, all_video_embeddings)
|
65 |
|
66 |
+
# Apply search context to the top 10 results
|
67 |
user_keywords = input_text.split() # Create a list of keywords from user input
|
68 |
+
weight = 1.0 # Base weight factor
|
69 |
|
70 |
for keyword in user_keywords:
|
71 |
if keyword.lower() in df["title"].str.lower().tolist(): # Check for matching keywords
|
72 |
weight += 0.1 # Increase weight for each match
|
73 |
|
74 |
+
# Adjust the recommendations based on the search context weight
|
75 |
+
final_recommendations = [
|
76 |
+
{key: value for key, value in video.items() if key != "embeddings"} # Exclude embeddings
|
77 |
+
for video in top_10_recommended_videos_features
|
78 |
+
]
|
|
|
|
|
79 |
|
80 |
+
# Apply the weight to sort the final recommendations (higher weight is better)
|
81 |
+
final_recommendations.sort(
|
82 |
+
key=lambda video: weight * dot_products[top_10_indices.index(video)], reverse=True
|
|
|
|
|
83 |
)
|
84 |
|
85 |
+
# Create the output JSON with the most similar video and final recommendations
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
86 |
output = {
|
87 |
"search_context": {
|
88 |
+
"input_text": input_text, # What the user provided
|
89 |
+
"weight": weight, # Weight based on search context
|
90 |
},
|
91 |
"most_similar_video": most_similar_video_features,
|
92 |
+
"final_recommendations": final_recommendations, # Top 10 with search context applied
|
93 |
}
|
94 |
|
95 |
return output
|
96 |
|
97 |
+
# Update the Gradio interface to output JSON with search context for the final recommendations
|
98 |
interface = gr.Interface(
|
99 |
fn=get_similar_and_recommend,
|
100 |
inputs=gr.Textbox(label="Enter Text to Find Most Similar Video"),
|
101 |
outputs=gr.JSON(),
|
102 |
title="Video Recommendation System with GNN-based Recommendations",
|
103 |
+
description="Enter text to find the most similar video and get top 10 recommended videos with search context applied after GNN-based search.",
|
104 |
)
|
105 |
|
106 |
interface.launch()
|