Identification of domain information for use in machine learning models

This invention describes a computer system that helps organizations identify important business terms from their documents to train artificial intelligence models. It works by first scanning all the organization's messy, unstructured text documents (like reports) to find potential key terms. Then, it cross-references these potential terms with more organized, semi-structured data (like spreadsheets or databases) to filter out irrelevant ones and confirm the truly important business-specific information. The final, refined list of key terms can then be used to build better AI.