Update app.py

app.py

@@ -1,54 +1,64 @@
+# app.py
 import streamlit as st
 from search_utils import SemanticSearch
 
 @st.cache_resource
 def init_search_system():
-    search_system = SemanticSearch(shard_dir="compressed_shards")
+    search_system = SemanticSearch()
     search_system.initialize_system()
     return search_system
 
-st.set_page_config(page_title="Research Paper Semantic Search with FAISS", layout="wide")
-search_system = init_search_system()
+def format_source(url):
+    """Convert URL to clickable link"""
+    return f'<a href="{url}" target="_blank" style="text-decoration: none;">🌐 Source</a>'
+
+st.set_page_config(
+    page_title="Semantic Search Engine",
+    page_icon="🔍",
+    layout="wide"
+)
 
-st.title("🔍 Research Paper Semantic Search")
-col1, col2 = st.columns([3, 1])
+search_system = init_search_system()
 
-with col1:
-    query = st.text_input("Search query:", placeholder="Enter your search...")
+# Custom CSS for better link display
+st.markdown("""
+<style>
+a.source-link {
+    color: #1a73e8 !important;
+    text-decoration: none !important;
+    transition: opacity 0.2s;
+}
+a.source-link:hover {
+    opacity: 0.7;
+    text-decoration: none !important;
+}
+</style>
+""", unsafe_allow_html=True)
 
-with col2:
-    top_k = st.slider("Results count:", 1, 20, 5)
-    threshold = st.slider("Similarity threshold:", 0.0, 1.0, 0.6)
+# Search interface
+st.title("🔍 Semantic Search Engine")
+query = st.text_input("Search knowledge base:", placeholder="Enter your question...")
 
 if query:
     with st.spinner("Searching through documents..."):
-        base_results = search_system.search(query, top_k)
-        threshold_results = search_system.search_with_threshold(query, top_k, threshold)
+        results = search_system.search(query, 5)
         
-    st.subheader("Top Matches")
-    st.dataframe(
-        base_results.style.format({'similarity': "{:.2%}"}),
-        use_container_width=True,
-        hide_index=True
-    )
-    
-    st.subheader(f"Filtered (>{threshold:.0%} similarity)")
-    if not threshold_results.empty:
-        st.dataframe(
-            threshold_results.style.format({'similarity': "{:.2%}"}),
-            use_container_width=True,
-            hide_index=True
+        # Format sources as links
+        results['source'] = results['source'].apply(
+            lambda x: f'<a class="source-link" href="{x}" target="_blank">🌐 View Source</a>'
         )
-    else:
-        st.warning("No results meet the similarity threshold")
-
-with st.sidebar:
-    st.header("System Info")
-    st.markdown(f"""
-    - **Loaded Shards:** {len(search_system.index_shards)}
-    - **Embedding Model:** `all-MiniLM-L6-v2`
-    """)
-    
-    if st.button("Clear Cache"):
-        st.cache_resource.clear()
-        st.rerun()
+        
+        # Display results with HTML
+        st.markdown("### Search Results")
+        for _, row in results.iterrows():
+            with st.container(border=True):
+                st.markdown(f"**{row['title']}**")
+                st.markdown(row['summary'])
+                st.markdown(row['source'], unsafe_allow_html=True)
+                st.progress(float(row['similarity']))
+                
+    # Sidebar information
+    with st.sidebar:
+        st.markdown("### System Status")
+        st.metric("Loaded Documents", f"{len(search_system.metadata_mgr.shard_map):,}")
+        st.metric("Index Shards", len(search_system.index_shards))