Artificial Intelligence is everywhere and it is impacting us in all areas of society and organizations. Although we could say that the greatest interest has been concentrated in AI based on Machine Learning and the analysis of large amounts of data (Big Data), a great opportunity is emerging with Artificial Intelligence based on contextual adaptation.
The type of AI that DARPA (Defense Advanced Research Projects Agency) considers the third wave of AI: Artificial Intelligence that allows to achieve understanding of the context through the representation of relationships, complexity and the meaning of data (semantics).
One of the main trends in semantic and contextual intelligence is knowledge graphs. The Gartner Hype Cycle for Emerging Technologies 2019 already considers it a technology that grows with expectations of significant impact in the short and medium term. Ricardo Alonso Maturana, founder of the Spanish company Gnoss, defines knowledge graphs as follows:
“A knowledge graph is a way of integrating and representing heterogeneous and distributed information, which allows discovering and investigating any topic in a more deep and intuitive and enjoy a more semantically web. A Knowledge Graph understands any fact about people, places and all kinds of things and how all these entities are connected to each other. It is a way to connect and unify information in a meaningful way and make it interrogated in a natural way for people in order to achieve a smarter web.”
As a simple, day-to-day example, we look at knowledge graphs in how Google searches work. Google collects and organizes millions of data about people, places and events to create meaningful and accurate interconnected search results.
In 2012 Google launched its knowledge graph as an extension of the search results that, in the form of a complement, is presented as a separate information box (see Figure 1). This box includes basic data, definitions of the search term, or secondary information.
Figure 1. Separate information box in Google searches.
Among some of the current uses of knowledge graphs we can mention the following:
Knowledge graphs have the ability to be used in data governance to centralize knowledge on “heterogeneous data sets” and constantly update them as more data is entered. Charts act as a semantic layer, modeling metadata, adding rich descriptive meaning to data elements.
A very successful example, and one that we can all see on its website, is the Prado Museum in Madrid. El Prado on the Web has integrated the existing data into various management systems: catalog of works, bibliographic records, information from the communication and marketing department; repositories; and specific projects from various departments. To do this, a unified and interrogable Knowledge Graph has been built under a Semantic Digital Model.
In this way, the documentation, editing, communication and publication processes are improved and a web view is generated as the first exploitation of the graph: www.museodelprado.es. (Information adapted from the website of Gnoss, the Prado Museum’s technology provider).
Representar escenarios de fraude de una manera gráfica visual, que es el núcleo de un gráfico de conocimiento, permite a los consultores financieros identificar para extender su trabajo de algoritmo de aprendizaje automático para considerar conjuntos de datos aún más heterogéneos que podrían no estar directamente relacionados con el tema en cuestión, o reconsiderar características y variables que las capacidades tradicionales de aprendizaje automático pueden ignorar. (Información adaptada de Medium.com y Analytics Vidhya).
Figure 2. Timeline as a representation of a knowledge graph in the Prado Museum web.
Finally, we can mention the applications that knowledge graphs can have in online education. From a knowledge graph, knowledge can be learned and shared, offering contextual information on the topic of interest. Navigation through knowledge graphs allows the learner to discover and investigate relevant information in a very intuitive way. Didactalia and Unikemia Upskilling are two examples of this application.
In conclusion, knowledge graphs are a trend with successful applications in the context of semantic and contextual intelligence. In future posts we will be delving a little more about this third wave.