Using unstructured data to create knowledge graphs in SurrealDB

The post explores the transformation of unstructured data into structured data using SurrealDB and LangChain, demonstrating how such data can be utilized for graph queries, vector searches, and data visualizations. It illustrates the process with a series of examples using SurrealQL and Rust, while guiding readers on how to recreate these steps in other languages like Python. The text explains the concept of unstructured data, its challenges for computer processing, and how recent advancements, particularly large language models (LLMs), have enabled the conversion of natural language into structured formats without human intervention. A key focus is on using a question-answering model from the Rust programming library to extract specific information from employee statements, like names, roles, and team affiliations, and structuring this data in SurrealDB for advanced querying and visualization. The post emphasizes using unique indexes and relation tables in SurrealDB to ensure data integrity, and demonstrates linking employees to their respective teams, managers, and companies. The article concludes by highlighting the ability to visualize this structured data graphically, promising further exploration in future posts.