Dieses IT-Projekt ist Teil unserer Digitalisierung und Optimierung der IT-Landschaft unserer Kunden. Durch gezielte Maßnahmen fördern wir den technologischen Fortschritt, optimieren systemübergreifende Prozesse und schaffen eine nachhaltige Basis für zukünftige Entwicklungen. Unsere IT-Referenzprojekte dienen als Grundlage zur Orientierung. Sie unterstützen die Wiederverwendbarkeit erprobter Konzepte im Rahmen der Projektumsetzung.
Dies ist eines von 4932 IT-Projekten, die wir erfolgreich mit unseren Kunden abgeschlossen haben.Wobei dürfen wir Sie unterstützen?
PTA Knowledge Graph
Kurzbeschreibung
Development of an AI-powered knowledge graph platform that automatically analyzes legacy systems and makes their business logic available for new development. Parsers convert source code, database structures, and requirements documents into a versioned knowledge graph, from which AI agents derive traceable and verifiable domain-specific descriptions. Domain experts review and supplement this content in a clarification workshop before it is used to create specifications, UI mockups, and handover packages for automated software development.
Ergänzung
The platform is a cloud-based reimplementation of a production-ready knowledge graph solution from a migration project. In the first iteration, deterministic parsers for Java and Oracle stored procedures are created; additional parsers will follow as needed. The parsers store versioned subgraphs as JSONL files in object storage. A pipeline validates references, IDs, and ontology. A central deployment API transfers the subgraphs to Neo4j, as well as chunks and embeddings to PostgreSQL with full-text and vector search. LLM agents derive domain-specific relationships and generate mockups and specifications. Hybrid retrieval, a state model, and approval gates support the process. The PTA transfers the concepts into the new architecture, implements parsers, the graph model, ontology, deployment interfaces, and agent processes, and ensures quality and traceability down to the source code line level.
Fachbeschreibung
The product supports the replacement of legacy systems that have evolved over time and are often inadequately documented, with their business logic primarily embedded in the code and the knowledge of a few experts. It makes business rules, processes, screens, and data flows visible in business terms and references each statement to its corresponding source. Domain experts specifically clarify open issues rather than completely redefining requirements from scratch, and approve business logic and mockups for each dialog. Migration occurs in distinct, business-defined phases, whose maturity level, coverage, and open issues can be measured at any time. Approved phases are handed over to automated development as verifiable specifications. After implementation, a comparison with the new code shows whether the target and actual states match. The benefits include reduced analysis effort, less reliance on individual experts, greater traceability, and a faster, lower-risk migration.
IT-Projektdaten
- Branchen Informations-Technologie
- Fachaufgaben KI: Allgemein, interfaces, Knowledge Management, Migration, Prototyping, Requirements Engineering, Reverse Engineering, software development
- DV-Aufgaben customised software new development
- Systemumgebungen Artificial Intelligence (AI), Cloud Computing
- Vorgehensmodell agile software development
- Programmiersprachen Kotlin, TypeScript
- Datenbanken Neo4J, PostgreSQL
- Entwicklungs- und Testtool GitHub, node.js
- Systemsoftware Docker
- Middleware und Frameworks Google Cloud Platform (GCP), vue js
Projektbeginn20.07.2026
Kunden, die uns vertrauen
Haben wir Ihr Interesse geweckt?
