A graduate NLP project analyzing whether supervised models could classify geopolitical evidence passages by analytical role and source perspective. Originally completed for IST 664 Natural Language Processing, the project later became the starting point for BLACKBONE™ Research on source-aware discourse graphs.
The project used a balanced geopolitical evidence corpus organized by event, source perspective, and analytical role. The main result was that source perspective separated more strongly than analytical role, while narrative, governance, leadership, and analysis labels often overlapped inside real geopolitical writing.
Final written report explaining the project design, modeling results, findings, and limitations.
Static PDF export of the final notebook submitted as the course project record.
The project later expanded into a source-aware graph research system under BLACKBONE™ Research.