Deriving knowledge from Knowledge Graphs
The only true wisdom is in knowing you know nothing.
The word ‘knowledge’ is a word used so often in our daily lives, without a universally accepted conceptual reference. The terms, ‘data and information’ and ‘information and knowledge’, are often used synonymously. However, as argued by Min Chen and David S. Ebert in their paper, ‘Data, Information, and Knowledge in Visualisation’ – data is not information and information is not knowledge. Philosophers and scientists have been struggling with the definition of the aforementioned terms for thousands of years, mainly because knowledge and information, in contrast to data, are the product of a synthesis in one’s mind and exist only in this mind.
But what if we map knowledge in an artificial mind?
We can then store all of this knowledge, free from human limitations, and use it based on our needs. This is the role ‘Knowledge Graphs’ have come to fulfil. Knowledge Graphs aim to reduce the keyword searching approach and instead, allow the user to focus on accessing the knowledge the data contains.
Before we analyse this further, let us first identify what is meant by ‘graph’, ‘knowledge’, ‘graph thinking’, ‘data’ and ‘information’.
What is a Graph?
A graph is a structure amounting to a set of objects in which some pairs of the objects are in some sense ‘related’. A graph is built by two elements – a vertex (or node) and an edge (or relationship). Each node represents an entity (any piece of data), and each relationship represents how two nodes are associated. This user-friendly structure enables graphs to model all kinds of different scenarios. And this flexibility makes them ideal for knowledge representation.
What are knowledge, information and raw data and how do they relate to one another?
‘Raw data’ put simply, are unprocessed records, and without context or interpretation, these records have no information. We know that ‘information’ provides answers to the who, what, where and when questions and is conveyed through the context of data and data combinations. Therefore, ‘information’ emerges through the processing of data shaped into a form that is meaningful to human beings. ‘Knowledge’ is what makes the transformation of ‘information’ into instructions, possible. It is the general understanding and consciousness gained from the accumulated information, meaning the experience is adjusted so that a new background or framework can be envisaged.
What is graph thinking?
By defining the above, we come to the conclusion that there is a relationship between ‘data’ (quality data to be precise) and ‘knowledge’. The realisation that there is value in understanding relationships across data, which can in turn lead to knowledge, is called ‘graph thinking’ – a term coined by Gosnell and Broecheler in 2020.
What is a Knowledge Graph?
A Knowledge Graph is a graph constructed by representing each entity as a vertex (or node), and then linking those nodes that interact with each other via edges (or relationships). These interlinked sets of data describe real-world entities, facts or things and their interrelations, in human understandable form. Unlike a simple knowledge base with a flat structure and static content, a Knowledge Graph acquires and integrates nearby information using data relationships to derive new knowledge. This ability to connect data and define relationships adds great value to users, applications, and even machine learning models.
Knowledge graphs empower the development of sophisticated AI. For example, it is impossible to train an autonomous vehicle for all possible light and weather conditions. However, it is possible to connect information from multiple contexts and infer the next action by using graph technologies.
Harnessing the power of Knowledge Graphs at Engine B
So, we now know that knowledge is created by information and information by data. This linear connection between data and knowledge is called graph thinking and a Knowledge Graph is a product of this reasoning. Representing knowledge in Knowledge Graphs eliminates the subjective factor of human nature, makes analysis more efficient, adds knowledge-value to users and applications and empowers machine learning models by extracting subgraph data. Undeniably, the Knowledge Graph has become a powerful asset, and this is exactly why it is being used by Engine B.
At Engine B, we believe Knowledge Graphs can greatly improve efficiency in professional services and many other industries. The work we are doing uses a combination of Machine Learning and Knowledge Graph Technology for projects such as fraud and anomaly detection – but in reality, the possibilities are limitless.