Imagine trying to find a specific fact in a library with a billion books, but many pages are missing or wrong. This project built a super-fast digital filing system and a smart assistant that doesn't just give an answer, but explains its reasoning through a conversation. It turns messy, giant webs of data into clear, trustworthy insights.
Large-scale knowledge graphs are often too slow to query, contain inconsistencies that crash standard engines, and provide 'black box' AI results that humans cannot trust or understand.
["Tentris: A high-efficiency graph database.", "A modular reasoner for detecting inconsistencies in billion-triple datasets.", "A retrieval-augmented generated explainer for co-constructive AI explanations."]
If you are a brand management agency dealing with fragmented customer data—this project developed a retrieval-augmented explainer that allows humans and machines to jointly produce understandable explanations for data-driven brand communication.
If you are a geospatial intelligence firm dealing with massive, inconsistent datasets—this project developed a modular reasoner that can handle over 1 billion triples to ensure data correctness and completeness.
If you are a software service provider dealing with slow data queries—this project implemented Tentris, a database that is up to 1000x faster than existing tools on large datasets like WikiData.
Based on available project data, specific pricing or licensing costs are not mentioned.
Yes, all approaches were evaluated on knowledge graphs with sizes of 1 billion triples or more.
Based on available project data, the specific IP and licensing terms are not provided.
It provides a unification theory for graph embeddings and a retrieval-augmented explainer to make machine learning results transparent and co-constructive.
The project period runs from 2022-10-01 to 2025-09-30, indicating the technology is currently in its final stages of development and evaluation.
The consortium consists of 6 partners across 5 countries, showing a strong European academic-industrial mix. With a 33% industry ratio (including 1 SME), the project balances deep theoretical research from universities with practical application requirements from commercial entities.
Contact Universitaet Paderborn regarding the Tentris database and explainable AI tools.
Contact us to bridge the gap between these research results and your corporate data strategy.
