Marble Releases Elementary Learning Concept Graph

Marble (Generative Spark, Inc.) has released Marble Skill Taxonomy v1, an open dataset that breaks down elementary curriculum concepts and connects them through prerequisite learning relationships.

• Organizes 1,590 micro-topics and 3,221 dependency relationships into a Directed Acyclic Graph (DAG)
• Classifies dependency relationships as hard or soft, and records the rationale for ordering in each relationship
• Provides each topic with a description, concept type, subject/domain, age range, and evidence and assessment questions for confirming mastery
• Includes 7 curricula, 3,261 achievement standards, and 1,859 pieces of linkage information

All data is provided in JSON format that can be used without any separate runtime or dependencies. Topics, dependency relationships, curriculum achievement standards, domain summaries, JSON schema, and checksums are organized in a GitHub repository.

Personalized learning systems or educational AI agents can use this as foundational data to recommend the next learning concept, track a learner's missing prerequisite knowledge, or map concepts across multiple curricula.