Create ingest/graph.py

ingest/graph.py

@@ -1,33 +1,29 @@
 import plotly.graph_objects as go
 
-
 def build_cluster_graph(results, clusters):
-    nodes_x = []
-    nodes_y = []
-    labels = []
+    xs, ys, labels = [], [], []
 
     i = 0
-    for agency, items in clusters.items():
-        for item in items:
-            nodes_x.append(i)
-            nodes_y.append(len(items))
-            labels.append(f"{agency}: {item['title']}")
+    for agency, docs in clusters.items():
+        for d in docs:
+            xs.append(i)
+            ys.append(len(docs))
+            labels.append(f"{agency}: {d['title']}")
             i += 1
 
     fig = go.Figure(
-        data=[
-            go.Scatter(
-                x=nodes_x,
-                y=nodes_y,
-                mode="markers",
-                text=labels,
-                marker=dict(size=10)
-            )
-        ],
-        layout=go.Layout(
-            title="Semantic Clustering of FOIA Documents",
-            xaxis_title="Document Index",
-            yaxis_title="Cluster Density"
+        go.Scatter(
+            x=xs,
+            y=ys,
+            mode="markers",
+            text=labels,
+            marker=dict(size=10),
         )
     )
+
+    fig.update_layout(
+        title="Semantic Cluster Graph (Click nodes to inspect documents)",
+        xaxis_title="Document Index",
+        yaxis_title="Cluster Density",
+    )
     return fig