Knowledge graphs are becoming essential for managing complex data in global supply chains, trade documentation, and logistics. However, many real-world knowledge graphs contain statements attributed to specific sources—for example, "according to Shipper A, container weight is 20 tons" or "according to Customs, container weight is 22 tons." Current provenance models can record who asserted what, but they treat provenance as semantically neutral, leaving underspecified how attributed claims relate to factual commitment, to one another, and to reasoning. This is a critical gap when decision-makers need to reconcile conflicting claims from different stakeholders.
According to a research paper titled "Provenance-Enhanced Statements in Knowledge Graphs" (arXiv:2606.15246), the authors—Fabio Vitali, Valentina Pasqual, and colleagues—propose a new framework called DEC. The paper notes that provenance-enhanced statements are "pervasive in contemporary knowledge graphs, especially in domains where graph content primarily represents claims, interpretations, and hypotheses (capta) rather than observer-independent facts (data)." Supply chain and trade are prime examples: bills of lading, customs declarations, and logistics visibility platforms are filled with attributed claims.
How DEC Works: Epistemic Stance and Cognitive Worlds
The DEC framework interprets provenance predicates as indicators of epistemic stance. Instead of treating provenance as neutral metadata, DEC uses cognitive modal logics (doxastic, epistemic, and conjectural) to characterize the attitude of the attributor toward the statement. It groups provenance-homogeneous sets of statements into cognitive worlds—each cognitive world represents a consistent set of claims from a particular source or perspective.
DEC defines a distinguished factual core called "reality". The framework characterizes locality, rationality, and controlled permeation between cognitive worlds and this factual core. This enables principled reasoning over attributed content without collapsing disagreements into inconsistencies. According to the paper, this allows "principled reasoning over attributed content without collapsing disagreements into inconsistencies." For example, conflicting weight claims from different parties can coexist in separate cognitive worlds, and reasoning can determine which world aligns with reality.
Technical Formulation and Implementation
The authors formalize a DEC interpretation for RDF datasets that is conservative over RDF 1.2 semantics. They clarify the role of intensionality and identity, including the classic Superman paradox (where different names refer to the same entity in different contexts). The approach is illustrated on common Semantic Web representations:
- Named graphs
- Quoted triples / RDF-star
- Reification
A prototype DEC reasoner has been implemented as a Fuseki dataset module. This module supports controlled factualisation—the process of elevating claims from cognitive worlds to reality—and explicit detection of disagreements and delusions (where a claim contradicts reality within a cognitive world).
Implications for Supply Chain and Trade Technology
For enterprise technology buyers, DEC offers a rigorous way to manage multi-source data in knowledge graphs. In trade finance, for instance, a letter of credit might reference documents from the importer, exporter, and bank—each with potentially conflicting statements. DEC can keep these assertions in separate cognitive worlds and reason about them without marking the graph as inconsistent. The framework also supports detection of disagreements, which could automatically flag discrepancies in customs filings or logistics track-and-trace data.
| Feature | Current Provenance Models | DEC Framework |
|---|---|---|
| Treatment of provenance | Semantically neutral | Epistemic stance indicator |
| Handling conflicting claims | Leads to inconsistency | Cognitive worlds with controlled permeation |
| Reasoning over attributed content | Limited | Principled, using cognitive modal logics |
| Prototype implementation | None cited | Fuseki dataset module |
According to the paper, DEC is particularly suited to domains "where graph content primarily represents claims, interpretations, and hypotheses." This directly maps to supply chain visibility platforms, customs technology, and trade finance systems where data comes from multiple, often contradictory, sources. The DEC reasoner prototype, built on Apache Fuseki, could eventually be integrated into existing RDF-based data lakes or knowledge graph platforms used by logistics tech vendors.
The research underscores the growing importance of provenance-aware reasoning as knowledge graphs become more central to digital trade and supply chain management. While DEC is still a research prototype, its formal grounding in logic and its conservative extension of RDF 1.2 semantics suggest a practical path toward adoption in enterprise systems. For technology leaders evaluating knowledge graph solutions, DEC represents a step toward more intelligent handling of attributed, often conflicting, information—critical for automating trade compliance, supply chain risk management, and multi-party document verification.