hyper-extract

Designing a YAML template for knowledge extraction involves defining entities, relations, fields, and identifiers to map unstructured text into strongly-typed schemas. This Skill provides prescriptive templates for record and graph extraction workflows, including model, list, set, hypergraph, temporal, and spatial formats.

Hypergraph extraction captures complex relationships where multiple entities interact simultaneously, going beyond standard binary graph relations. You need it when analyzing interconnected events like corporate ownership structures or battle scenarios where standard graph templates cannot represent multi-party relationships accurately.

Yes, YAML templates can extract temporal and spatial data from historical texts using specialized template designs. The Skill includes curated cases like battle_analysis and biography_events, enabling structured extraction of time-bound and location-specific knowledge from unstructured historical documents.

Yes, the template design workflow supports multilingual descriptions for extracted entities and display labels. It guides users through optimizing naming conventions and generating multilingual content, ensuring knowledge artifacts are accessible across different language contexts.

The best way to validate YAML schema correctness is using the built-in validation steps in the template workflow. This Skill includes optional optimization and validation phases that check YAML structure, verify entity-relation mappings, and ensure strongly-typed outputs meet schema requirements before deployment.

Available YAML extraction template types for financial data include model, list, and set templates for structured records. The Skill provides curated cases like earnings_summary, enabling extraction of company names, revenue figures, and reporting periods from unstructured financial reports into validated schemas.