Automated Citation Network Analysis for Research Papers

Create a Python script using NetworkX and Pandas that can parse a large corpus of scientific PDFs to construct a citation network graph. The script should extract citation metadata, build a directed graph representing academic paper relationships, and calculate network metrics like centrality, clustering coefficient, and influence scores. Include functionality to visualize the network using Plotly and generate a comprehensive JSON report of key network characteristics, with special attention to handling complex citation formats across different scientific domains.