Lost in translation: data integration tools meet the Semantic Web (experiences from the Ondex project)

More information is now being published in machine processable form on the web and, as de-facto distributed knowledge bases are materializing, partly encouraged by the vision of the Semantic Web, the focus is shifting from the publication of this information to its consumption. Platforms for data integration, visualization and analysis that are based on a graph representation of information appear first candidates to be consumers of web-based information that is readily expressible as graphs. The question is whether the adoption of these platforms to information available on the Semantic Web requires some adaptation of their data structures and semantics. Ondex is a network-based data integration, analysis and visualization platform which has been developed in a Life Sciences context. A number of features, including semantic annotation via ontologies and an attention to provenance and evidence, make this an ideal candidate to consume Semantic Web information, as well as a prototype for the application of network analysis tools in this context. By analyzing the Ondex data structure and its usage, we have found a set of discrepancies and errors arising from the semantic mismatch between a procedural approach to network analysis and the implications of a web-based representation of information. We report in the paper on the simple methodology that we have adopted to conduct such analysis, and on issues that we have found which may be relevant for a range of similar platforms

💡 Research Summary

The paper investigates how graph‑based data integration and analysis platforms, exemplified by the Ondex system, can consume information published on the Semantic Web. Ondex, originally designed for systems biology, represents knowledge as a graph of concepts (nodes) and relations (edges), enriched with ontology‑derived types, attributes, provenance, and evidence. Although this structure superficially resembles RDF triples, the authors uncover several semantic mismatches that hinder straightforward mapping to Semantic Web standards.

A central case study is the “CV” (Controlled Vocabulary) field. In practice, CV is used both as a namespace for identifiers (e.g., GO, UniGene) and as a provenance marker indicating the source database or format (e.g., ATRegNet, NWB). This dual usage leads to ambiguous URIs when converting Ondex data to RDF, because the same CV value may represent different semantic roles. The authors systematically catalog this behavior by inspecting parser plugins and observing how CV values are assigned in various contexts.

Another key issue is identifier scope. Ondex generates internal integer IDs for concepts when a graph is imported, and links external identifiers (Accessions) to these IDs. This document‑centric approach assumes an implicit scope defined by the source file, which is unsuitable for the distributed, globally addressable environment of the Semantic Web where URIs must provide unambiguous, global identifiers. Consequently, the implicit scoping in Ondex conflicts with the explicit, web‑wide scoping required by RDF.

To address these problems, the authors propose a six‑step methodology: (1) enumerate all data‑model elements (Concept, Relation, ConceptClass, RelationType, Generalized Data Set, CV, Accession, EvidenceType, Context); (2) define intended semantics based on documentation and stakeholder interviews; (3) observe actual usage through code inspection and expert analysis; (4) reconcile observed usage with intended definitions, producing a refined semantic model; (5) formulate best‑practice guidelines that can be applied without code changes; and (6) suggest future development actions to enforce the guidelines.

Applying this methodology to CV, they recommend splitting CV into two distinct elements: “Namespace” (to be used only with identifiers and mapped to a specific URI) and “Provenance” (to capture the last authoritative source of a piece of information). They advise avoiding vague CV values that refer to families of ontologies or to technology platforms, and they propose that each (Namespace, Accession) pair be semantically equivalent to a single URI. For provenance, the CV should denote the most specific authority responsible for the data, not the processing tool or institution.

The paper further discusses broader implications for other network analysis tools such as Cytoscape, Neo4j, and Gremlin‑based systems. It highlights common pitfalls: conflating knowledge external to the system with analysis‑derived metrics within the same graph, and mixing different information bases (e.g., raw data, derived scores, layout coordinates) without explicit typing. These practices are acceptable in procedural, document‑centric workflows but become problematic when data are to be shared on the open web.

By providing a concrete, repeatable assessment process and concrete recommendations, the authors demonstrate how to align graph‑based platforms with Semantic Web principles, ensuring consistent identifier scoping, clear provenance tracking, and unambiguous mapping to RDF/OWL. Their work serves as a blueprint for developers seeking to make network analysis tools interoperable with distributed knowledge bases, ultimately fostering more robust, reusable, and semantically sound bioinformatics infrastructures.