Ontologies allow us to capture domain-specific knowledge. For instance, how do you teach a machine what a contract is? What is a receipt? What is a customer? All of these questions deal with the nature of people, places, or things.

Without understanding domain-specific concepts, machine learning often produces results that are not directly applicable to a business domain. This means you need a team of SMEs to interpret and validate the output of a complex model and determine its fitness for purpose. Using ontologies, we provide machine learning with the relevant business context to produce meaningful results.

In traditional data and analytics stacks, there is much emphasis on information extraction using any number of machine learning techniques, from classical statistical methods to modern deep learning algorithms. However, there is no explicit effort on capturing system or domain-level knowledge.

Our approach enables us to build data science solutions that implicitly use domain knowledge to derive more relevant and meaningful models that know how to interpret the data they are using to make predictions.

case studies