Upload app.py

src/app.py

@@ -780,7 +780,8 @@ with tab_kg_exp:
 
         n_seeds = st.slider("Number of seed papers", 3, 15, 6, key="kg_exp_n_seeds")
 
-        EDGES_PER_TYPE = 10   # 각 edge type당 고정 샘플 수 → 10 types × 10 = 최대 100 edges
+        # edge type당 고정 샘플 수 — 10 types × 10 = 최대 100 edges
+        EDGES_PER_TYPE = 10
 
         with st.spinner("Querying graph..."):
             top_seeds = (kg_nodes_exp[kg_nodes_exp["node_type"] == "seed_paper"]
@@ -791,8 +792,8 @@ with tab_kg_exp:
             if seed_ids:
                 ids_sql = ", ".join(f"'{sid}'" for sid in seed_ids)
 
-                # 1-hop: seed paper에 연결된 모든 edge type (journal/author/affiliation/
-                #         city/country/field/citation_event 포함)
+                # 1-hop: seed paper에 연결된 모든 edge type
+                # → journal / author / affiliation / city / country / field / citation_event
                 hop1 = _ddb.execute(f"""
                     WITH ranked AS (
                         SELECT source, target, edge_type,
@@ -806,15 +807,20 @@ with tab_kg_exp:
                     WHERE rn <= {EDGES_PER_TYPE}
                 """).df()
 
-                # 2-hop: citation_event → citing_paper / intent
-                event_ids = [
-                    x for x in
-                    set(hop1["source"].tolist()) | set(hop1["target"].tolist())
-                    if str(x).startswith("event:")
-                ][:30]
-
-                if event_ids:
-                    ev_sql = ", ".join(f"'{eid}'" for eid in event_ids)
+                # 2-hop: kg_nodes_exp 타입 기반으로 citation_event 노드 ID 추출
+                # (prefix 가정 없이 실제 node_type 컬럼으로 확인)
+                hop1_all_ids = set(hop1["source"].tolist()) | set(hop1["target"].tolist())
+                event_node_ids = (
+                    kg_nodes_exp[
+                        kg_nodes_exp["node_id"].isin(hop1_all_ids) &
+                        (kg_nodes_exp["node_type"] == "citation_event")
+                    ]["node_id"].tolist()[:40]
+                )
+
+                if event_node_ids:
+                    ev_sql = ", ".join(f"'{eid}'" for eid in event_node_ids)
+                    # citation_event → HAS_CITING_PAPER → citing_paper
+                    # citation_event → HAS_PRIMARY_INTENT → intent
                     hop2 = _ddb.execute(f"""
                         WITH ranked AS (
                             SELECT source, target, edge_type,
@@ -823,7 +829,11 @@ with tab_kg_exp:
                                    ) AS rn
                             FROM read_parquet('{kg_edges_path}')
                             WHERE (source IN ({ev_sql}) OR target IN ({ev_sql}))
-                              AND edge_type IN ('HAS_CITING_PAPER','HAS_PRIMARY_INTENT')
+                              AND edge_type NOT IN (
+                                  SELECT DISTINCT edge_type
+                                  FROM read_parquet('{kg_edges_path}')
+                                  WHERE source IN ({ids_sql}) OR target IN ({ids_sql})
+                              )
                         )
                         SELECT source, target, edge_type FROM ranked
                         WHERE rn <= {EDGES_PER_TYPE}
@@ -867,7 +877,7 @@ with tab_geo:
         fig_map.update_layout(geo=dict(showframe=False), height=500)
         st.plotly_chart(fig_map, use_container_width=True)
 
-    st.subheader("Top Cities (Affiliation)")
+    st.subheader("Top Cities")
     city_cnt = (seed_filtered.merge(
                     aff_geo_df[["affiliation_name","city_name","country_name"]],
                     left_on="affiliation", right_on="affiliation_name", how="left")
@@ -881,16 +891,46 @@ with tab_geo:
             .update_layout(xaxis_title="", yaxis_title="# Seed Papers", xaxis_tickangle=-40),
             use_container_width=True)
 
-    st.subheader("Citation Trend over Time (selected paper)")
-    trend2 = (seed_events.dropna(subset=["citing_year"])
-              .assign(citing_year=lambda df: df["citing_year"].astype(int))
-              .groupby("citing_year").size().reset_index(name="count"))
-    if not trend2.empty:
-        st.plotly_chart(
-            px.line(trend2, x="citing_year", y="count", markers=True,
-                    title="Citations per Year")
-            .update_layout(xaxis_title="Year", yaxis_title="Citations"),
-            use_container_width=True)
+    # ── Affiliation 시각화 ──────────────────────────────────────
+    st.subheader("Top Affiliations")
+    geo_col1, geo_col2 = st.columns(2)
+
+    with geo_col1:
+        aff_cnt = (seed_filtered[seed_filtered["affiliation"].str.strip() != ""]
+                   .groupby("affiliation").size()
+                   .reset_index(name="count")
+                   .sort_values("count", ascending=False).head(20))
+        if not aff_cnt.empty:
+            st.plotly_chart(
+                px.bar(aff_cnt, x="count", y="affiliation", orientation="h",
+                       title="Top 20 Affiliations by Seed Papers",
+                       labels={"count": "Seed Papers", "affiliation": ""})
+                .update_layout(yaxis=dict(autorange="reversed"),
+                               xaxis_title="Seed Papers", yaxis_title="", height=520),
+                use_container_width=True)
+
+    with geo_col2:
+        aff_country = (seed_filtered[
+                (seed_filtered["affiliation"].str.strip() != "") &
+                (seed_filtered["country"].str.strip() != "")
+            ]
+            .groupby(["country", "affiliation"]).size()
+            .reset_index(name="count")
+            .sort_values("count", ascending=False)
+        )
+        top_affs = aff_country.groupby("affiliation")["count"].sum().nlargest(20).index
+        aff_country_top = aff_country[aff_country["affiliation"].isin(top_affs)]
+        if not aff_country_top.empty:
+            st.plotly_chart(
+                px.bar(aff_country_top, x="count", y="affiliation",
+                       color="country", orientation="h",
+                       title="Top Affiliations by Country",
+                       labels={"count": "Seed Papers", "affiliation": "", "country": "Country"})
+                .update_layout(yaxis=dict(autorange="reversed"),
+                               barmode="stack",
+                               xaxis_title="Seed Papers", yaxis_title="",
+                               legend_title="Country", height=520),
+                use_container_width=True)
 
 
 # ═══ 7. ANALYTICS ═══════════════════════════════════════════════
@@ -1030,6 +1070,24 @@ with tab_analytics:
                 use_container_width=True,
             )
 
+    # ── Citation Trend over Time ────────────────────────────────
+    st.markdown("---")
+    st.subheader("Citation Trend over Time (selected paper)")
+    st.caption("How citations to the selected seed paper have changed year by year")
+    trend_sel = (seed_events.dropna(subset=["citing_year"])
+                 .assign(citing_year=lambda df: df["citing_year"].astype(int))
+                 .query("citing_year >= 2000")
+                 .groupby("citing_year").size().reset_index(name="count"))
+    if not trend_sel.empty:
+        st.plotly_chart(
+            px.line(trend_sel, x="citing_year", y="count", markers=True,
+                    labels={"citing_year": "Year", "count": "Citations"})
+            .update_layout(xaxis_title="Year", yaxis_title="Citations",
+                           hovermode="x unified"),
+            use_container_width=True)
+    else:
+        st.info("No citation trend data for the selected paper.")
+
     # ── Export ─────────────────────────────────────────────────
     st.markdown("---")
     st.subheader("Export Data")