Maya Natarajan at graph database specialist, Neo4j describes how to use Knowledge Graph-based digital twin technology for better supply chain management.
More and more chief information officers (CIOs) see potential in digital twins. Digital twins are creating virtual, highly detailed, and faithful reproductions in software of real-world assets, such as a factory or parts of cities.
Many CIOs, however, struggle with optimising complex supply chains especially at the moment. Our world, already battered by a pandemic, faces geopolitical tensions that place huge stresses on the tightly interconnected but fragile supply chain. The good news is that technology in the form of knowledge graphs and digital twin technologies can provide insights into supply chain optimisation.
Knowledge graphs have been around for almost 50 years. They mainly languished in the academic world until, in 2012, Google announced it was using a knowledge graph behind its search engine. Since then, the convergence of analytics, data science, machine learning, and artificial intelligence (AI) has generated the need for knowledge graphs. This is because they have the capability to make each of these technologies better, smarter, and more predictive.
At its most basic, a knowledge graph is an interconnected dataset enriched with meaning or semantics. It allows its users to reason about the underlying data and use it for complex decision-making.
Knowledge graphs are effective because of their graph database schema. The reason is down to the inherent limitations of SQL and relational when it comes to supporting queries.
It’s also down to the specific shape of the data you’re going to want to work with in a supply chain context.
From big data to small and wide data?
Data in today’s world comes in various shapes and sizes, but CIOs have found that the data that holds the greatest insight is never simple to work with. The data that contains the greatest insights is complex, connected, and can be deeply hierarchical and recursive. In fact, 99 times out of 100, it’s hidden. Prior to the pandemic, we used to talk about Big Data. Now, the move is toward what Gartner terms ‘Small and wide’ data. Small and wide data provides more context, especially for machine learning programs, and this is what we need to address.
The fact is Small and Wide cannot be analysed with traditional technologies. When such complex data gets ingested into a property graph store, relationships between data are coded in. These relationships provide the first level of context to data. In a graph, this ‘dynamic context’ means the graph grows and gets increasingly rich as new information is dynamically added.
For a knowledge graph, reading the relationship from storage and querying the graph is as straightforward as traversing a graph. Teams can add a third element, semantics, to get afull knowledge graph, as well as algorithms and other tools. The fact is it’s really easy with graph technology to create a rich, reactive representation of complexity, like a digital twin of a supply chain.
The pandemic has exacerbated the lack of visibility of supply chains, but it is also hard to gain complete visibility into a supply chain because they are complex multi-dimensional connected digital networks. As such, they can only really be modeled as a knowledge graph. That’s the best tool to connect all its facets from materials to products, plants to distribution centres, and shipping. The knowledge graph provides the context so decisions can be made holistically, taking the many interlocking dependencies into consideration.
The digital twin knowledge graphs for better supply chain management bring data together and creates a connected virtual supply chain. It allows managers to better organise, analyse, and visualise their data. Managers get a trackable but also in-depth picture of all the products, suppliers and facilities in that supply chain, and the relationships between them.
Power plant management
Tata Consulting’s TCS IP2 power plant management SaaS service has been so enhanced by adoption of a digital twin service fueled by a knowledge graph that it has helped its customers achieve a 9% reduction in emissions, lower fuel utilisation, and US$6 million (€5.47 million) annual operational savings. That’s just one example.
There are other examples of knowledge graphs making equally tangible bottom line uplift. Putting a supply chain or manufacturing digital twin into a graph gives you real-world fidelity in everything from the oil and gas sector to nationwide retail distribution.
Complex data is naturally modelled as a knowledge graph
From logistics and operations, all the way to marketing sales and services, digital twin war gaming of complex supply chain use cases is already happening.
If you’re building a supply chain digital twin, you should model it as a knowledge graph. That’s because the complex data supply chain you want to capture is naturally and a lot more easily modeled as a knowledge graph. And to seal the deal, a knowledge graph provides the flexibility, performance, and analytical capabilities needed to build, manage and query digital twins at enterprise scale.
The author is Maya Natarajan, senior director, product marketing, at native graph database specialists Neo4j.