Semantic Product Lifecycle Management with Knowledge Graphs

Posted by Boris Shalumov on Dec 20, 2021 8:00:00 AM
Find me on:

Unleashing the power of knowledge is nowadays the most crucial task for enterprises in order to stay competitive. However, knowledge is not just data thrown into a database. It is a complex, dynamic model that puts every piece of information into a larger frame, builds a world around it and shows its connections and meaning in a specific context.

At Cambridge Semantics we’ve been building sustainable solutions that dissolve data silos and enable enterprises to think as a collective. This is accomplished via Anzo, our standards-based knowledge graph overlay that can connect all your structured and unstructured data sources, providing levels of insight and strategic decision advantages that otherwise would not be possible.

All that said, let’s jump into this use case about semantic Product Lifecycle Management (PLM) and how knowledge graphs can help enable it. Product structures are multidimensional hierarchies by nature. Product life cycles are usually split up by phases and disciplines, which naturally create silos. 

Due to mass production and strong dynamics the automotive industry is one of the most complex manufacturing related industries in terms of product data management. Compound this with the fact most car manufacturers do not offer only one single model to their clients. All models must be strongly intertwined as they share parts, delivery trucks and manufacturing lines.

This demo shows you how we integrate said data throughout the product lifecycle, which:

  • Provides transparency over the so-called product DNA, 
  • Reduces time to market, 
  • Configuration management, 
  • Enabling design to line workflows

and much more. 

This demo is a quick overview of all these use cases, but I think it will give you a pretty good idea of how PLM with knowledge graphs can work. All right, so let's watch the demo:


To get started Anzo will guide you through the whole workflow of creating a Knowledge Graph for PLM. 

There are 4 stages:

  1. Onboarding: where you get all your data sources. 
  2. Modelling: where you create your canonical model and the target ontology.
  3. Blending: where you connect all your data and map it to the target ontology (this is the heart of Anzo).
  4. Access: where you build your local framework-based applications or dashboards on top of the knowledge graph. 

In Anzo using the example PLM Graphmart, we have an explore tab. This provides a great visualization that explains what is actually going on in the graphmart. The first item of note, is the different contributing data sources. 

In our PLM example, most of the sources deal with sales figures, or PLM data or suppliers, and the canonical model looks quite simple. The canonical model has different car configurations, consisting of parts organized in the hierarchical nature and where the suppliers are that supply these parts. 

You can also see how and why the connections were made in the model using the Anzo Ontology Editor. There you’ll find the data profiles and attributes in table format, to help users get a good grasp of their data and sources relationships. 

Moving on to the Anzo Data Layers tab, you can load data right from all the different sources. In our example it’s SAP and Teamcenter among other data sources. The interesting part is connecting all the data. Anzo automatically transformed the sales data to the target ontology, the supply chain management, the bomb data. This is done by Anzo automatically writing queries that take the source data and map it to the target ontology. It can be done manually, but you can also use the Anzo automated query features of exact match, fuzzy match, and regex match. 

In the demo I use fuzzy match at a 20% threshold to find similarities between the different objects and different sources. It generates a new layer, which suggests three different matches with varying degrees of relevance. With the matches found, Anzo supplies a SPARQL query for easy mapping to the target ontology.

In the third layer of my example you’ll find a pre-calculation of some product structure KPIs like PageRank calculations. It estimates how important the part is for the whole product structure based on its connections to other parts; and if the product changed, how that would affect the manufacturing line. These graph algorithms are easy to use and we continue to strive to make them easier. 

Now, if we look at Access in Anzo, you’ll find a dashboard that lets us analyze the criticality of any one supplier. This is done based on the count of a part and configurations touched, plus how many different parts the Supplier supplies. 

This is all built just by clicking, aka traversing the knowledge graph and selecting relationships of interest. Building tables in Anzo is easy. You simply start by clicking a path, then clicking different relationships in the graph. This allows business users to easily interact with the Knowledge Graph without really needing to know how to work the underlying magic. 

In our example, it’s so fun and exciting to me how easy it is to identify the dependency one might have on a supplier. Answering questions like:

  • What happens if the contracts change?
  • What is the supplier goes bankrupt?
  • How many configurations will be affected?
  • How will this affect configurations sold?

Is really easy to find these answers and make decisions for your future. When doing this easy point and click analysis, Anzo also automatically generates relatable filters to make it as easy as possible to hone in on your question with any stipulations. In the background, Anzo is automatically generating SPARQL queries.

For further product structure transparency, I’d like to also highlight the capabilities of Anzo Network Navigator. It’s easy to navigate through data points, relationships, sources; and the built-in visualizations make it easy to identify risk type questions. In our example, questions like which parts shouldn't be touched or changed, and what the consequences would be of a change.

You can also easily identify patterns, for example cycles. For example, seeing hierarchy issues is glaringly obvious. In the same family, you can’t have a child be the parent of a grandparent. In graph, it’s easy to identify that error and further analyze what’s happening.

Anzo Network Navigator makes it easy to create transparent, flexible dashboards from all these different sources allowing users to make better decisions. 

I hope you found this demo helpful and inspirational as you start brainstorming how and where knowledge graphs can be most valuable in your organization. Please feel free to reach out with any questions you may have

If you’re interested in exploring more knowledge graph use case demonstrations, I recommend watching our recent Knowledge Graph Demo Showcase.





Tags: Data Management, Data Integration, Anzo, Graph Database, Knowledge Graph, semantic graph, RDF

Subscribe to the Smart Data Blog!

Comment on this Blogpost!