Unlocking 25 Years of R&D data through graph visualization
The outcomes
- A comprehensive data warehouse supporting advanced analytics and reporting.
- A graph structure, enabling visual exploration of material histories
- Researchers generate insights faster, without IT technical expertise.
The context
A multinational oil and gas company operates an R&D hub in Belgium focused on advanced materials, recycling, and biofuels. Over 25 years, the site accumulated vast datasets—covering input materials, outputs, chemical properties, and analyses. By 2020, however, limited in-house data expertise made accessing and leveraging this information for research and business intelligence a challenge.
The logic
The company needed a way to transform decades of siloed data into actionable insights for scientists and data teams. Traditional relational databases were cumbersome for researchers, especially chemists unfamiliar with complex query languages. To accelerate innovation, the data had to be accessible, intuitive, and capable of revealing hidden relationships.
The solution
We first consolidated the historical data into a robust data warehouse, completed in 2023, enabling structured analysis and reporting. To make this data truly user-friendly, we introduced a Neo4J-based graph application that visually represents relationships between materials and processes. Researchers can now explore sample histories and connections through intuitive graph views—without writing queries.
To further enhance usability, we integrated a Large Language Model (LLM) to translate natural language questions into graph queries, and implemented graph embedding algorithms to predict potential links and similarities between nodes.
Next steps in applied excellence
These enhancements will significantly increase the value of the R&D dataset, enabling deeper insights and accelerating innovation in plastics, recycling and biofuels.