The Transparent Relations Ontology (TRO): a vocabulary to describe conflicts of interest

TL;DR

The paper tackles the fragmentation of data about conflicts of interest by introducing the Transparent Relations Ontology (TRO), a lightweight OWL vocabulary designed for minimal modelling and interoperability. TRO enables publication of data about time-bound roles in entities, backed by explicit evidences, and reuses established vocabularies (e.g., GIST, GoodRelations, PPROC, ePO, Schema, Time, DBpedia) to support interoperability. A concrete BCITRG use case demonstrates how TRO can drive a RDF knowledge graph that links Basque tender data with investigative journalism to surface potential conflicts of interest. The work emphasizes rigorous ontology development practices (CI, ROBOT, OQuaRE) and FAIR-aligned publication with persistent URIs, aiming to improve transparency and data reuse, with planned future enhancements including more tests and broader adoption across domains.

Abstract

The Transparent Relations Ontology (TRO) offers a vocabulary to publish data about relations between powerful parties that should be more transparent, in order to detect possible conflicts of interest. TRO is based on minimal modelling, reusing common vocabularies to offer a simple yet useful resource to publish interoperable data about pointers to relations that might result in corruption cases. Additionally, best practices have been followed in order to sustain a technically rigorous ontology development process. A usage example with real data is mentioned, integrating information from Basque Government's Open Data services and a news outlet. Building upon its foundational design, future enhancements of TRO could significantly amplify its utility in uncovering and scrutinizing opaque relationships that may lead to corruption.

Figures (3)

• Figure 1: Basic modelling behind the TRO ontology: a person has a role during a certain time in a given organization, and such fact is backed by an evidence. The orange ovoids represent OWL classes; the arrows represent OWL restrictions; dashed boxes represent XSD Data Types. Not all the classes nor all the properties are shown: for all the entities of the ontology refer to Figure \ref{['fig:entities']}.
• Figure 2: Entities of TRO, as shown in the Protégé interface. Hierarchies represent class/subclass relations and property/subproperty relations. Left: classes; center: object properties; right: data properties.
• Figure 3: Basque Country Institutions Transparent Relations Graph (BCITRG). The data is obtained from two sources (top): the Basque Government's tender registry, and a news source. The data is transformed to an RDF KG (Grey line encircling ovoids and arrows). A simplified sample is shown, and URIs are reduced for readability. Orange ovoids represent TRO classes; grey ovoids represent RDF nodes; dashed boxes represent literal values; TRO properties are used in all the RDF triples. The role of the person is identified by a unique URI composed by the person's normalised name, role type, dates of the role, and the organization, obtaining a completely unique URI for each role, on each date, on each organization, backed by an specific evidence.