Welcome to Knowledge Graphs. In this article, we will introduce a sort of tools which is being increasingly used in digital transformation initiatives. These tools take information management to a new level and dramatically increase our ability to interact with large volumes of data.
What is a Knowledge Graph?
A Knowledge Graph is a structured way of organising and representing information, typically used to show the relationships between different entities or concepts. It looks like a kind of network that connects different pieces of information, by nodes which represent entities (such as people, places, companies, etc.) and edges, which describe the relationships between them. All these pieces together form knowledge about a given area or domain, which is also structured and easily processed by a machine.
In simple terms, it is as if our data can be better "understood" thanks to their relationships, allowing us to go beyond the simple search for information by also exploring the context that surrounds it.
As the number of nodes and relationships in the graph grows (it can consist of millions of them), it is very likely that there are certain patterns in the way entities are related, which are not visible to the naked eye, but well exploited are useful for different situations (recommendation, anomaly detection, inferring new relationships).
Practical Example: Google Knowledge Graph
If you have ever searched on Google and seen a panel with additional information on the topic, you have experienced the power of a Knowledge Graph. This approach transforms the search experience, taking us from just text strings to extended concepts. When you search for "the white house" you not only get a list of pages that mention it, but also a summary with key facts, such as location, architects and more, and enabling to click and explore in detail those data to discover new interesting information.
What benefits do they offer?
1. Context: Knowledge networks offer a contextualised view of data, representing information in a way that reflects the real world. This facilitates decision-making processes by providing a holistic view of connected information.
For example, let us suppose a knowledge graph that represents what an engineering organisation has learned in the execution of its projects. This graph would relate concepts such as projects and lessons learned, and also concepts related to the management of each project (people involved, meetings, milestones, actions, problems, risks) and to the products generated (requirements, technologies, tests, etc.). With this type of networks, each lesson learned would have the context of the problems that originated it, meetings that took place, participants, actions, and decisions, and take everything into account during the planning phase of new projects with similar characteristics.
Example of the context of a Lesson Learned within a project.
2. Flexible Data Integration: For those responsible for information management in organisations, networks are connected data warehouses that provide enormous flexibility to integrate data from various sources. They simplify the centralisation, standardisation and governance of data to ensure reliability and efficiency.
3. Intuitive Access for Non-Technicians: Even for those without deep technical knowledge, such as business analysts or specialists, representing information through graphs provides a visual and intuitive interface. Navigating and following relationships becomes an effective way to understand the whole context.
4. Improved data for AI applications: Because of the way they organise information, they represent an improved data source for their application in multiple fields of AI: improving the models generated by Machine Learning processes, inference of new information from existing information, or support for the development of intelligent systems capable of learning and adapting. Currently, they are seen as the ideal complement for the generation of LLM's (for applications such as ChatGPT), providing them with a source of contrasted information that avoids the "hallucination" problems they usually suffer from.
Knowledge Graphs are a key tool to consider for effective digital transformation. Their ability to connect data, facilitate collaboration and provide a deep understanding of relationships makes them an essential tool for any organisation looking to optimise its management processes.
At Immedia, we currently use knowledge graphs as part of our solutions within the space domain, particularly in the fields of project management, systems engineering (particularly model based engineering or MBSE) and, of course, in our knowledge management solutions.
In future articles, we will explore in depth the implementation, storage and processing capabilities of Knowledge Graphs.