Break Data Silos with Knowledge Graphs
The idea of eliminating data silos is not a new one. For years, employees and executives have cried out for The One Solution to save them all – one that gives them access to hard-to-find information that only exists in the hard-to-access land of Another Department. When this data is locked in silos, what’s the solution?
For years, thought leaders have talked about “Breaking Silos” in hundreds of LinkedIn posts, HBR articles, and blog posts (like this one), claiming that it will solve all the problems in your org.
Let’s take Customer Service as an example. Many executives think that Customer 360 is the way to better customer service. Your consulting partners have probably told you that attain customer 360 vision, you need to eliminate data and system silos accumulated over the years. Since these silos have caused a lot of headaches and missed opportunities within organizations, the idea of ripping them out and replacing them with one magical customer view sounds like the answer to all your problems.
However, removing silos is not realistic – the cost alone is immense and the effort of managing change is overwhelming. Data Lakes and Data Warehouses have not materialized as the solution either.
There’s a better way to get you deep context for each department – and it involves establishing connections between different entities in your company’s data.
We can do that using a knowledge graph. And to make it scalable, we’ll use an Ontology.
Wait, what’s a Knowledge Graph?
Here’s a quick definition from Wikipedia: “A knowledge graph formally represents semantics by describing entities and their relationships”
I still don’t know what a knowledge graph is
Think of Knowledge Graphs as a way establish the relationships between things, such as people, or places or dates or objects. Let’s consider the movie Forrest Gump as an example. We know that:
- Tom Hanks and Sally fields both acted in Forrest Gump
- The dialog was in English
- Forrest Gump was released in 1994
- You can watch it on Apple TV or Amazon (but you have to rent or buy it).
We can represent this relationship in a knowledge graph, where the circles represent objects, and the connecting lines represent a relationship.
Okay, what’s an Ontology?
An ontology is a type of data model (schema) that defines relationships between real-world entities such as objects, events, situations, or abstract concepts.
Ontology + Data = Knowledge Graph
We can use the Forrest Gump knowledge graph to build a Movie Ontology that can apply to all movies.
This can apply to an enterprise too. Let’s consider Utilities – the knowledge graph that connects entities could include objects like a customer, his apartment, a meter, and a bill. We can extend it with location information and environmental objects like weather and traffic. Here’s what a simple Utilities Ontology can look like –
Since this information exists in your enterprise software (like SAP ERP and CRM) and databases (like SAP HANA), we can very quickly create a knowledge graph of your Utility operations. We can rapidly turn silos of data in your Utilities applications and databases into one connected knowledge graph without much development effort.
App Orchid has built one of the first Utilities Ontologies as the base of our AI Platform. We can use your enterprise data and our ontology to create a unique knowledge graph of your business. This lets us apply our Machine Learning and AI tools to get insights incredibly quickly!
Bottom line – if you’re really interested in getting the value of breaking silos, you want to use Knowledge graphs.
- Establishing relationships between data means that Knowledge Graphs effectively break data silos, but without breaking your systems, or replacing them,
- Once again, no system replacement. Our knowledge graphs augment your systems, processed and people.
- Knowledge graphs let you perform complex operations on them such as combinatorial analysis and machine learning to identify relationships, correlations, patterns, and trends.
- Knowledge graphs help you make sense of unstructured data such as text in documents or service notes. Consider a contract – most contracts are hard enough for humans to read. Knowledge Graphs can extract and report on key data from them, such Contract Parties, Renewal Dates and Liabilities.