Using historical books to create structured knowledge graphs in SurrealDB

The blog post focuses on using historical texts to create structured knowledge graphs in SurrealDB, employing advanced natural language processing models like SentimentModel and ZeroShotClassificationModel to analyze sentiment and interactions between countries during significant historical periods, specifically around World War I. The project involves breaking down historical books into manageable chunks, extracting text from PDFs, and utilizing Rust-based sentiment analysis to evaluate changes in public perception towards various European nations across different years. The process involves setting up a SurrealDB environment, running sentiment and interaction analysis on text passages, and visualizing the results through graph visualizations to discern patterns and trends. The analysis highlights the evolving sentiment towards countries like Germany, France, and Belgium, providing insights into historical narratives and international relations during and after the war. The blog also discusses methods to improve the accuracy of sentiment and interaction detection and suggests that the approach can be scaled to handle larger datasets for more comprehensive historical research.