Deployed Research System·May 2026 – Present·Universität Leipzig

DLT Knowledge DSS & Cross-Marketplace Trust

Structuring reusable DLT design knowledge into an Open Knowledge Format (OKF) and operationalizing it for cross-marketplace identity and trust.

Cover image for DLT Knowledge DSS & Cross-Marketplace Trust
Confidentiality Note: Unpublished manuscripts and proprietary evaluation benchmarks withheld pending formal peer review.

Overview & Research Motivation

Design-science knowledge for distributed ledger systems is fragmented across hundreds of academic papers, industry whitepapers, and technical specifications, making it difficult to reuse systematically. At the Information Systems Institute (Universität Leipzig), I developed a researcher-facing decision-support system (DSS) for structuring, querying, and synthesizing DLT design knowledge.

The workbench replaces opaque vector embeddings with a native Open Knowledge Format (OKF), graph-structured concept representation, deterministic multi-stage lexical and structural retrieval, and provenance-bound context for evidence-grounded synthesis. I then operationalized these design principles to design a decentralized product registry addressing listing fragmentation, counterfeit risk, and review state across e-commerce marketplaces using GS1 Digital Link, GTIN, and ECLASS standards.

The Core Systems Problem

  • Standard vector-only RAG approaches discard critical structural relationships, hierarchical taxonomy, and citation provenance, hallucinating architectural recommendations for distributed systems.
  • In global e-commerce, identical physical products exist under fragmented, incompatible marketplace representations with isolated seller reputations, untraceable supply chains, and duplicate product data.
Formulated Question: "How can design knowledge for distributed architectures be represented with explicit provenance, and how can that knowledge guide verifiable state exchange across competing e-commerce marketplaces?"

Why I Worked on It

"I wanted to test whether we could capture nuanced architectural design science without reducing academic literature to disjoint text chunks, and immediately prove the utility of that knowledge on a difficult real-world information systems challenge."

My Specific Technical Contributions

  • Architected the Open Knowledge Format (OKF) schema with multi-tier concept hierarchies and explicit architectural relationship mappings.
  • Designed and deployed the deterministic graph retrieval engine combining lexical search, typed relationship traversals, and citation tracking.
  • Formulated the cross-marketplace product registry architecture utilizing GS1 Digital Link, GTIN identifiers, and ECLASS classifications.
  • Built the interactive researcher-facing exploration UI deployed live at dsr-library.org.
Architecture: Open Knowledge Format (OKF) & Graph-Grounded Decision Support
1. DSR Literature Ingestion
Open Knowledge Format · Explicit Taxonomy
Structured concept hierarchies
2. Deterministic Graph Traversal
Hybrid Lexical + Concept Neighborhood Search
Zero ungrounded hallucinations
3. E-Commerce Trust Layer
GS1 Digital Link · GTIN · ECLASS Registry
Cross-marketplace verifiable claims

System Architecture & Verification Pipeline

OKF Knowledge Representation

Decomposes DLT literature into structured design artifacts, architectural trade-offs, formal evaluation models, and boundary conditions with explicit entity-relationship graphs.

Deterministic Hybrid Retrieval

Executes typed graph queries paired with BM25 lexical ranking and concept-neighbourhood traversals, ensuring zero ungrounded context injection.

Cross-Marketplace Trust Registry

Decentralized state layer anchoring verifiable claims for product specifications, verified reviews, and merchant credentials mapped via GS1 Digital Link.

Design Decisions & Engineering Trade-Offs

Explicit graph relationships over vector-only similarity
Rationale: Vector embeddings lose causal links between architectural choices and their trade-offs.
Trade-Off: Requires structured extraction schemas, but guarantees verifiable evidence grounding.
GS1 Digital Link URI standard for product identification
Rationale: Seamlessly bridges physical barcode scanning with distributed web-based resolver infrastructure.
Trade-Off: Must map multiple legacy marketplace proprietary schemas (Amazon ASIN, eBay ePID).

Technical Stack & Tools

TypeScriptNext.jsPythonNeo4j / Graph DBGS1 StandardsTailwind CSS

Known Limitations

  • Current ontology extraction requires manual validation for edge-case DLT consensus mechanisms.
  • Cross-marketplace smart contract settlement layer evaluated on test environments; full mainnet deployment pending pilot partner onboarding.

What I Would Test Next

  • Expanding automated ingestion pipelines with self-supervised verification checks against canonical standards.
  • Evaluating multi-party dispute resolution protocols on simulated adversarial marketplace state transitions.

Connected Systems & Inquiries