import pandas as pd
import networkx as nx
from pyvis.network import Network
import ast
from itertools import combinations
import os
def generate_graph():
print("Loading references.csv...")
try:
df = pd.read_csv("references.csv")
except FileNotFoundError:
print("references.csv not found.")
return
print(f"Processing {len(df)} papers...")
# Initialize Graph
G = nx.Graph()
# Process Authors
for index, row in df.iterrows():
try:
# Authors are stored as string representation of list "['Author A', 'Author B']"
# ast.literal_eval is safer than eval
authors = ast.literal_eval(row['authors'])
# Add nodes
for author in authors:
if not G.has_node(author):
G.add_node(author, title=author, group=1)
# Add edges (co-authorship)
if len(authors) > 1:
for a1, a2 in combinations(authors, 2):
if G.has_edge(a1, a2):
G[a1][a2]['weight'] += 1
G[a1][a2]['value'] += 1 # Pyvis uses 'value' for edge thickness
else:
G.add_edge(a1, a2, weight=1, value=1)
except (ValueError, SyntaxError) as e:
# Handle cases where author string might be malformed
continue
print(f"Graph stats: {G.number_of_nodes()} authors, {G.number_of_edges()} collaborations.")
# Visualize
net = Network(height="750px", width="100%", bgcolor="#222222", font_color="white")
net.from_nx(G)
# Set physics layout suitable for large graphs
net.barnes_hut(
gravity=-8000,
central_gravity=0.3,
spring_length=150,
spring_strength=0.001,
damping=0.09,
overlap=0
)
output_file = "author_network.html"
net.save_graph(output_file)
print(f"✅ Created interactive graph: {os.path.abspath(output_file)}")
if __name__ == "__main__":
generate_graph()
