Knowledge graphs: Finding what I need

Back in 1945, Vannevar Bush criticized the completely outdated tools used in knowledge management at the time in his legendary essay "As we may think". Almost 80 years later, we are still hearing the same complaints - digital age or not.

For decades, knowledge was recorded, linked and disseminated in databases. The technical reproduction of knowledge was exorbitantly accelerated as a result. WhaNo Code at itt has not been promoted, however, are creative knowledge processes that generate new knowledge by linking existing information in new ways.

The dream of data disciples that new knowledge would always be gained if only GPUs were fast enough has since been shattered. Not because the idea is fundamentally bad, but because the data to learn what we want to learn simply does not exist.

For this reason, knowledge graphs have always been at the heart of all Empolis applications, because they combine two essential properties:

First, they offer the ability to capture and map complex semantic relationships based on just a few sample data sets.

Secondly, they allow different users with different perspectives to build and maintain a common information base and use it as a common basis for very different decisions.

Most experts who use knowledge graphs as a special form of database technology have not yet penetrated to their revolutionary core. Used intelligently, knowledge graphs allow us to capture, process and use information the way people think. For this to work, however, the possibilities of semantically linked information must be accessible to users. It is not enough for software developers to simply replace their old backend with a knowledge graph. Instead, we need to enable our users to work together on linked data. Each from their own perspective and with their own specific business needs, but all in the same knowledge network.

Find what I need. Not what I'm looking for.

Let's look at the simple example of searching for information: As an employee of a technology company, there can be many reasons to search for a scanning electron microscope : Perhaps you simply want to know what such an instrument is needed for. Or you urgently need one and would like to know if there is one in your company that you can use. Or maybe you are already sitting in front of the microscope and want to look something up in the user manual. You may also need technical support from a microtechnologist for a project. There are different sources for each piece of information - with different purposes, different information models and different ways of searching for it. And even after years in a company, you will not know all these sources.

The revolutionary thing about knowledge graphs is that they link all these sources of information together in such a way that they know which interest inscanning electron microscopes can be satisfied from which of these sources. At the same time, knowledge graphs know which features they can use to distinguish between the different types of information. On this basis, they can do something very human - namely ask back: "What exactly interests you about a scanning electron microscope? Would you like to use one? Do you need documentation or are you looking for an expert?" In this way, the user is easily guided through a complex data landscape to the information they need. Even if the specific search query has been made for the first time.

One, two, twin

Knowledge graphs also offer great added value in other areas: many companies are working hard to offer their customers an optimal digital experience. They are faced with the challenge of capturing everything a good sales representative (or application consultant or service expert) knows about the product and its interdependencies (e.g. which parts are compatible, which configuration offers which functions) in digital form. In this digital representation, the product data must be linked to the customer's needs. Today's customers expect solutions to their problems - not functions or performance data.

Knowledge graphs must be accessible both to product experts who maintain knowledge and to software services that operate with it. This single source of truth for all channels - we call it thedigital information twin - is the basis for smart suggestions in service and maintenance or in the selection and configuration of products. Capturing product knowledge, including all interdependencies, requires a knowledge graph that is powerful but also flexible and maps complex relationships in a natural way.

Linking metadata instead of software development

In recent years, whenever market researchers have conjured up hype cycles for new technologies, an enormous effort has been made to keep up with the trend. This is different with knowledge graphs - if they are used intelligently. On the one hand, this is because knowledge graphs have a long history and have already been tested and matured in many business-critical solutions. On the other hand, knowledge graphs are simply real team players.

Their aim is not to replace existing technologies and infrastructures. Rather, it is about using existing metadata to bring together existing collections of information, databases and documents. With minimal set-up costs, knowledge graphs facilitate access to corporate knowledge - for example through a central knowledge management portal, as we implement in our cloud solutions for our customers in the technical services and chemical industries. And the best thing is: applications based on knowledge graphs learn by linking existing data correctly, so that users are constantly developing their solutions - simply by using them. This is "No Code at it."

And since existing information systems are usually already secured by an existing rights management system, the access rights can be derived from the connections in the knowledge graph. This means that everyone can only see what they are allowed to see without any additional effort. Simply, quickly and in one central location.

Why the hype? And why now? Why is this such a trending topic?

Knowledge graphs work the way people think. This allows people to work seamlessly with AI and expand their problem-solving capabilities. Today, this is no longer an abstract vision of the future. Knowledge graphs create value in leading industry applications:

Empolis Service^Express® leverages the search and metadata integration capabilities of Knowledge Graphs to provide service technicians with exactly the information they need to solve their specific service problem - giving them superhuman powers.

Empolis Content ^Express® leverages the capabilities of Knowledge Graphs to map a digital twin and match product features with customer needs as they select, configure and use the product - creating a superhuman product experience.

Empolis Knowledge ^Express® leverages the intelligent search, matching and integration capabilities of Knowledge Graphs to support sophisticated tasks such as market and technology research, risk assessment or investigative tasks - turning its users into superhuman knowledge workers.

What's behind all this?

By combining the advantages of RDF-based semantic networks and property graphs, the Empolis Knowledge Graph offers an expressive power that goes beyond that of other knowledge graph systems. A reliable but flexible schema, multiple hierarchies and inheritance of relations and attributes as well as the ability to freely link everything - schemas and data, objects, attributes and relations are core competencies when it comes to capturing complex knowledge, e.g. in the field of digital information.

The Empolis Knowledge Graph puts the power in the hands of business experts - not developers. Using graphical user interfaces, they can understand and edit the Knowledge Graph, customize the model, create questions and rules, and synchronize with external data sources. With the Empolis Knowledge Graph, they can even participate in the development of applications and services.

The Empolis Knowledge Graph is a market-ready solution for mission-critical applications. This goes beyond the patented access rights system and ranges from tracking dependencies in the semantic model, live backup and schema transfer between environments to support for cloud and on-premise solutions and many other services that support productive use at an industrial level.