Update app.py
Browse files
app.py
CHANGED
@@ -19,6 +19,12 @@ def load_data():
|
|
19 |
print(f"Embedding shape: {embeddings.shape}")
|
20 |
print(f"Number of patent numbers: {len(patent_numbers)}")
|
21 |
print(f"Number of metadata entries: {len(metadata)}")
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
return embeddings, patent_numbers, metadata
|
23 |
|
24 |
embeddings, patent_numbers, metadata = load_data()
|
@@ -63,7 +69,14 @@ def search(query, top_k=5):
|
|
63 |
continue
|
64 |
patent_data = metadata[patent_number]
|
65 |
result = f"Patent Number: {patent_number}\n"
|
66 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
67 |
result += f"Similarity Score: {1 - distances[0][i]:.4f}\n\n"
|
68 |
results.append(result)
|
69 |
|
|
|
19 |
print(f"Embedding shape: {embeddings.shape}")
|
20 |
print(f"Number of patent numbers: {len(patent_numbers)}")
|
21 |
print(f"Number of metadata entries: {len(metadata)}")
|
22 |
+
|
23 |
+
# Print sample metadata
|
24 |
+
sample_patent = next(iter(metadata))
|
25 |
+
print(f"Sample metadata for patent {sample_patent}:")
|
26 |
+
print(json.dumps(metadata[sample_patent], indent=2))
|
27 |
+
|
28 |
return embeddings, patent_numbers, metadata
|
29 |
|
30 |
embeddings, patent_numbers, metadata = load_data()
|
|
|
69 |
continue
|
70 |
patent_data = metadata[patent_number]
|
71 |
result = f"Patent Number: {patent_number}\n"
|
72 |
+
|
73 |
+
# Safely extract abstract
|
74 |
+
abstract = patent_data.get('abstract', 'No abstract available')
|
75 |
+
if isinstance(abstract, str):
|
76 |
+
result += f"Abstract: {abstract[:200]}...\n"
|
77 |
+
else:
|
78 |
+
result += f"Abstract: Unable to display (type: {type(abstract)})\n"
|
79 |
+
|
80 |
result += f"Similarity Score: {1 - distances[0][i]:.4f}\n\n"
|
81 |
results.append(result)
|
82 |
|