Intelligent Situation Handling with the Knowledge Graph
In this blog post, I want to introduce you to the Situation Knowledge Graph – it’s the basis of the Explore Related Situations app, which became available with the November release of Intelligent Situation Automation.
To see how you can use the Situation Knowledge Graph to understand a situation better, let’s take a look at these situation templates:
It looks like they represent three independent situations. However, if they sometimes occur at the same time, it’s an indication that you might have a more serious issue in your business. For example, there could be a problem with a specific material.
For every single situation we know what the respective primary affected business entities are. We will also retrieve the data context, which is other business data related to the situation. If we look at a “Stock Transport Order Overdue” situation, the affected entities are the purchase order and the specific purchase order item, the goods issue material document, and the goods receipt material document. The data context can consist of more business data elements, such as the material number. This information is needed for simple rule-based automation. However, it is not sufficient for a deeper understanding of the situation and the improvements to the business processes in the future.
We don’t have any information at hand about how a situation relates to other situations, business applications, processes, solutions, APIs, and so on. It’s not clear which relationship the individual business entities have with each other. It’s also hard to understand where a situation is placed in the greater enterprise model. We also don’t know about the platform and available tools and services. There is also no information from outside the core business process application or third-party solution providers.
However, all this information is out there. It just resides in many different systems, is distributed all over the IT landscape, coded in different formats, and accessible via different mechanisms and protocols, if it’s accessible at all. There would be so many more possibilities if we could access it easily and use it for situation handling.
The Situation Knowledge Graph is here to help solve that problem.
What is a Knowledge Graph?
The term “Knowledge Graph” first appeared in 2012 in a blog post by Google’s Amit Singhal ”Introducing the Knowledge Graph: things, not strings“. He describes a “Knowledge Graph” as something that helps you understand the relationships between things. It is a made-up phrase without a clear definition, inspired by the visualization as a line graph. The term was widely adopted to denote a database that describes things and their relationships uniformly as nodes and edges.
Situation Knowledge Graph
We build a knowledge graph by extracting, analyzing, and transforming a wide range of metadata, which is available in the various systems at SAP. We use metadata from our development systems such as
- S/4HANA for the virtual data model,
- Business Situation Type API for situation master data,
- FIORI Apps Reference Library about applications,
- API Business Hub about publicly available APIs, CDS views, and Events,
- Extensibility Cockpit about business contexts,
and many other internal systems.
All this metadata is harmonized, unified, and tied together to then form one consistent and deeply connected knowledge graph based on a semantic data model. Information that resides in silos behind different systems, protocols, and data formats is now easily accessible via the knowledge graph query interface.
Explore Related Situations
You can drill down to explore the problem with the materials. On the second page, you can now see which specific materials are causing so many situations. In this example, it’s “TG11”, which has the highest number of situations in four different situation templates. However, there is also “FLOG-SP14-NO-QM”, which has an even higher number of situations but occurs only in three situation templates.
You decide to take a closer look at “TG11” and select this line. You will now see a projection of the Situation Knowledge Graph related to this material and the most relevant related entities. In the blue box, for example, the related entity “Material Group” is shown, while the thickness of lines in the box indicates that most situations are with material group “L001“. You select another related business entity type for further analysis.
On the left-hand side of the diagram, you can see more details from the related situation templates and data context, as well as which applications are related to them and are used to deal with the situation. The Situation Knowledge Graph page has a lot of interactive elements, so you can view it from different angles and navigate freely to other related business entities.
In the future, we hope we can build more applications that take advantage of the Situation Knowledge Graph and to provide a new user experience. Besides UI applications, we also see a lot of potential for utilizing the knowledge graph for recommendation services, which can be embedded into every application.
Want to know more?
Try Intelligent Situation Automation For Free: Get Started