Skip to main content
Knowledge Graph is a structured collection of datasets that connects academic standards, curricula, and learning science data. By bringing high-quality datasets together, enriching them, and unifying the schema, Knowledge Graph enables edtech developers to focus on building instructional tools and AI-powered experiences that can improve teaching and learning. Knowledge Graph data can be accessed in multiple ways, including downloadable files, APIs, and MCP-based integrations. We are continuing to expand these access patterns in collaboration with close partners and welcome feedback on what would be most useful for your workflows. Our downloadable datasets are provided in JSONL format, a graph-native format that preserves the structure of entities and relationships. This makes it straightforward to work with the data across graph databases, relational systems, in-memory tools, and AI pipelines without requiring specialized infrastructure.

​
What you can do with this data

  • Standards alignment: Identify how your content supports specific academic standards and create content rooted in learner competencies across all key subjects.
  • Instructional planning: Create dependencies, learning progressions, and content coverage, starting with math in the Common Core State Standards.
  • Compare state standards: Adapt content aligned to one state standard to other states, initially in math across the Common Core State Standards and 15+ additional states.
  • Curriculum alignment : Align your content or create additional materials aligned to the curriculum. To support these use cases, Knowledge Graph organizes educational information into a small set of structured datasets. Contact support@learningcommons.org ↗ for information about access to gated functionality like this.

​
How to access it

​
Datasets in Knowledge Graph

Knowledge Graph organizes educational information into structured datasets. These fall into four main groups:
Dataset TypeWhat It RepresentsExamples
CurriculumLessons, activities, materials, and assessments from publishers.Illustrative Mathematics 360 ↗
Academic standardsHierarchical structures of state or national learning goals.State standards from CASE Network 2 ↗
Learning progressionsLogical and usually sequential ordering of learning targets.Student Achievement Partners’ Coherence map ↗
Learning componentsAcademic standards broken down into concrete skills or concepts.ANet’s learning component for K12 math ↗

​
How the data works

Datasets are modeled as graphs using two complementary components: entities (nodes) and relationships (edges). Together, these define both the elements in the dataset and the connections between them.
  • Entities (nodes): Each node represents a distinct concept in the domain, such as a standard or learning component, and is identified by a UUID. These nodes define the core elements of the dataset.
  • Relationships (edges): Each edge represents a directed connection between two nodes, expressed as a triple (source UUID → relationship type → target UUID). These edges define how entities relate to one another, forming the graph’s structure. Conceptually, this is similar to a join table in a relational database, but instead of foreign keys, UUIDs are used to link nodes directly.
Every entity and relationship carries a UUID. Some UUIDs come from external sources such as the CASE Network ↗, which helps ensure interoperability. UUIDs make it easy to join files, reference entities, and preserve links across datasets. The consistent use of UUIDs across files allows deterministic joins and avoids tight coupling to any one database system. You can think of these UUIDs as “graph-friendly join keys” that make connections explicit without relying on schema-level constraints. Knowledge Graph is designed to be database-neutral, lightweight, and interoperable. The files mirror the structure of a graph model, with entities as nodes and relationships as edges, while remaining easy to ingest in common environments. This design supports both deterministic joins and AI workflows, such as embeddings or RAG. To get started, check out the tutorials that walk through typical use cases and how to slice data to meet your needs.