Spaces:
Sleeping
Sleeping
Create ingest/graph.py
Browse files- ingest/graph.py +33 -0
ingest/graph.py
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import plotly.graph_objects as go
|
| 2 |
+
|
| 3 |
+
|
| 4 |
+
def build_cluster_graph(results, clusters):
|
| 5 |
+
nodes_x = []
|
| 6 |
+
nodes_y = []
|
| 7 |
+
labels = []
|
| 8 |
+
|
| 9 |
+
i = 0
|
| 10 |
+
for agency, items in clusters.items():
|
| 11 |
+
for item in items:
|
| 12 |
+
nodes_x.append(i)
|
| 13 |
+
nodes_y.append(len(items))
|
| 14 |
+
labels.append(f"{agency}: {item['title']}")
|
| 15 |
+
i += 1
|
| 16 |
+
|
| 17 |
+
fig = go.Figure(
|
| 18 |
+
data=[
|
| 19 |
+
go.Scatter(
|
| 20 |
+
x=nodes_x,
|
| 21 |
+
y=nodes_y,
|
| 22 |
+
mode="markers",
|
| 23 |
+
text=labels,
|
| 24 |
+
marker=dict(size=10)
|
| 25 |
+
)
|
| 26 |
+
],
|
| 27 |
+
layout=go.Layout(
|
| 28 |
+
title="Semantic Clustering of FOIA Documents",
|
| 29 |
+
xaxis_title="Document Index",
|
| 30 |
+
yaxis_title="Cluster Density"
|
| 31 |
+
)
|
| 32 |
+
)
|
| 33 |
+
return fig
|