This repository contains resources, data, and tooling for building and evaluating domain-specific knowledge graphs within the HACID project.
The HACID project focuses on developing hybrid human–AI collective intelligence systems for decision support in complex, open-ended domains such as medical diagnostics and climate services. A core component of this approach is the construction of case knowledge graphs (CKGs) and domain knowledge graphs (DKGs), which integrate heterogeneous data sources into a structured, semantically rich representation.
This repository provides materials related to the creation, population, and validation of such knowledge graphs.
- Ontology definitions – schemas and vocabularies used to model domain knowledge
- Data sources – structured and semi-structured datasets used to populate the graph
- RDF generation – scripts and pipelines for transforming data into RDF triples
- Evaluation artifacts – test cases for assessing the ontologies wrt competency questions
The knowledge graphs in this project rely on Semantic web standards (e.g., RDF, OWL), build on modular and reusable ontology components, and are developed using a combination of:
- Top-down ontology engineering and design patterns
- Bottom-up data integration from multiple sources
These approaches ensure consistency, interoperability, and scalability across domains.
The ontologies folder contains the core semantic models used to define and structure HACID knowledge graphs.
This folder includes:
-
Core / upper-level ontologies – High-level, reusable concepts and properties that provide a common semantic foundation across the knowledge graph.
-
Cross-domain foundational modules – Modular ontology components that capture specific subdomains or knowledge areas, designed to be reused and combined into larger ontology networks.
-
Domain ontologies – Specialized models describing application domains (medical diagnostics and climate services), defining domain-specific entities, relationships, and constraints.