OBIE

Description:

• The Ontology Based Inference Engine (OBIE) is a simple reasoning engine that performs inference and entailments based on logic encoded in formal ontologies. At present, this is limited to the propagation of knowledge annotated to a specific term or class in a hierarchical ontology to a parent class in the hierarchy.

• For example, when a source reports that "Metformin treats Type I Diabetes Mellitus", OBIE will logically entail that "Metformin treats Diabetes Mellitus" (the parent class of Type I Diabetes Mellitus in the MONDO disease ontology that supports the data) - and instantiate this inferred statement/edge.

• More formally:

  • Premise 1: "Metformin treats Type I Diabetes Mellitus"
  • Premise 2: "Type I Diabetes Mellitus subclass of Diabetes Mellitus"
  • Entailment: "Metformin treats Diabetes Mellitus"

• This is a valid ontological entailment under a 'some-some' interpretation of knowledge assertions - i.e. asserting that Metformin treats a class/category of diseases like 'Diabetes Mellitus' means that there are at least some cases in which this drug can treat some instance or form of the disease (not that the drug treats every form or instance of the disease).

• This type of ontology-based knowledge propagation is important for retrieving relevant results for queries of data where knowledge is annotated to concepts at different levels of an underlying ontology. For example, without OBIE, queries for drugs that treat Diabetes Mellitus would not return Metformin if this assertion were only explicitly reported at the level of 'Type I Diabetes Mellitus'.

In the future, the inferences OBIE performs may expand to cover other types of ontology-based entailments like knowledge propagation across partonomic chains encoded in ontologies (e.g. if Gene X is expressed in Tissue Y, and Tissue Y is part of Tissue Z, then Gene X is also expressed in Tissue Z).