Related Ontologies and Models

The SEPIO development process included a landscape analysis of existing models covering evidence and provenance related concepts, and evaluating them against the use cases defined above. Here we aimed to understand if existing work could be re-used or extended to meet the needs of our driving projects.

The Provenance Ontology (PROV-O) is a generic model about provenance of any type of entity, describing entities it derived from, through what processes, and through the contributions of what agents. While it lacks coverage of evidence-related concepts, we are exploring how we might use or extend the model in our representation of provenance for claims and data supporting them.

The Evidence and Conclusion Ontology (ECO) provides an account of evidence, but only as summary level codes that indicate a type of evidence used to support a claim. ECO is used to annotate claims, but it is not meant to be used as a model to structure accounts of evidence and provenance. We are exploring the use of ECO classes to type instances of evidence lines as defined in SEPIO, to establish a connection with this widely used resource.

Efforts such as the Ontology of Biomedical Investigations (OBI), the Investigation-Study-Assay (ISA) model, and Knowledge Engineering from Experimental Design (KEfED) framework offer accounts of experiments and observations that produce data used as evidence, but do not model the space between this data and claims it is used as evidence to support. We are exploring how to align with these efforts in our representation of data-generating processes.

The Semantic EvidencE (SEE) Framework and Micropublication model consider evidence in the context of scientific argument and explanation. They have been developed and applied primarily toward curating the literature to extract claim networks and structured logical arguments, and thus are not suited to efficiently address our use cases around evidence and provenance as found in curated databases. SEPIO is well aligned conceptually with the SEE framework in particular, and we are exploring how these efforts may complement and align with each other to span their respective domains of application.

The Open Biomedical Annotations (OBAN) model is a small ontology with similar scope to SEPIO, but lacks the depth and distinctions needed to meet our core use cases. We do, however, leverage the OBAN ‘association’ model in RDF representation of claims in our driving projects.