Most of the papers, collected in DB of articles created in ProOptiBeef, have proved and disproved statements attached. The statements were extracted from the articles by a group of experts who analyzed the papers within ProOptiBeef project. OntoBeef was applied to represent them. For this purpose we have created a new ontology, we called it OntoBeef Science, which imports OntoBeef Domain and uses its classes to express scientific laws.

Currently Science provides the taxonomy of scientific laws and its ontological characterization.

The types of scientific laws, we adopted in the project, were taken from the works of Polish philosophers of science and nature: K. Ajdukiewicz and W. Krajewski. They represent standard, classical views on science which we find appropriate for our task as confluent with the usual practice in the kind of science we deal with. The distinctions proposed by them were analyzed in terms of their use in ontological modeling and reasoning. Then we have chosen the types of laws which are present in our domain of interest and specified their meaning.

Figure 1 depicts the final taxonomy of scientific laws.

We use OWL-API application to deduce new scientific laws on the basis of ones already coded in Science ontology. We shall point out only two of possibly many ways of obtaining new knowledge from our ontology.

The first type of reasoning is illustrated in figure below.

It concerns functional laws. Having two laws in the ontology, e.g. law 1 treating Q1 as dependent quality and Q2 as independent one, and having law 2 treating Q2 as independent quality and Q2 as dependent one, the reasoning system generates a new scientific law (law 3) treating Q1 as dependent quality and Q3 as independent one.

Example: From two theses present in Science:

– rate of glycolysis ⊑ ∃ isIndependentParameterIn {t7}

– beef aging time ⊑ ∃ isDependentParameterIn {t7} and

– beef aging time ⊑ ∃ isIndependentParameterIn {t4}

– beef tenderness ⊑ ∃ isDependentParameterIn {t4}

our application infers a law:
– rate of glycolysis ⊑ ∃ isIndependentParameterIn {t-new}

– beef tenderness ⊑ ∃ isDependentParameterIn {t-new}

New laws found by the application are intended to be verified by the domain experts and if accepted, will be added to the list of laws.

The second type of reasoning is a ‘reverse inheritance’ of scientific laws. It is illustrated in figure below.

Let us consider class C1 with two children C2 and C3 and assume that C1 is not governed by any scientific law in our ontology, whereas C2 and C3 are governed by scientific laws: 1 and 2 respectively. In this situation our system signalizes that although there is no law governing all instances of C1, there are laws which take into account some of them. System points out the subclass of C1 and provides information about the laws governing them. For instance, “sensory attribute” has no law in our prototype, whereas its subclass “beef tenderness” or “beef color” are governed by laws. While browsing “sensory attribute” system informs about the laws of its subclasses.

Comments are closed.