The document discusses using Neo4j and graph databases to analyze documents from the Panama Papers investigation. It describes the steps taken by journalists to acquire, classify, and analyze over 3 million documents. Key aspects included developing a data model to represent entities and relationships, extracting metadata and parsing documents to populate the graph. Neo4j was highlighted as a tool to efficiently store and query the connected relational data in the documents. An example graph showing relationships in documents related to Azerbaijan's president was also presented.