Case study WOWnet – Intelligent connectivity platform for goods and know-how
Challenge
Companies face growing complexity in matching supply and demand within rapidly changing markets. Traditional manual methods were too slow and ineffective for analysing massive, non-homogeneous data sets needed to form strategic partnerships. Unite turned to fortiss to research AI approaches to automate and optimise knowledge extraction and business relationship modelling.
Solution
fortiss developed advanced AI methods to effectively represent and extract information from large, diverse datasets for Unite. They focused on graph neural networks and semantic models to capture relationships between entities in business contexts. Novel approaches such as Metapath- and Entity-aware Graph Neural Networks (PEAGNN) and Variational Embeddings for Community Detection and Node Representation (VECoDeR) were developed to learn meaningful representations for recommendation systems and community detection. These methodologies enabled automated, precise matching of business needs and capabilities, facilitating agile and targeted partnerships within complex networks.
Result
- Developed Metapath- and Entity-aware Graph Neural Network (PEAGNN) for improved recommendations
- Created VECoDeR model for joint node representation and community detection
- Enabled automated, scalable extraction and structuring of business-relevant knowledge from Big Data
- Facilitated faster and more accurate matching of supply-demand relationships across industries
- Demonstrated that advanced graph AI can transform traditional business networking approaches
Outcome
The WOWnet project shows that integrating AI-based graph modelling and knowledge representation can revolutionise how companies form strategic relationships. fortiss’ innovative approaches not only improved recommendation accuracy and community detection but also proved scalable and robust, providing a foundation for future AI-driven business connectivity platforms in dynamic markets.
Project partner
