Knowledge engineering has a relatively long history in AI, dating back to early expert systems. It is often considered in contrast to more data-centric approaches, including machine learning, as the more "traditional" approach to AI: A method based more on symbolic representation, expertise and the manual gathering of knowledge, fundamentally different from, possibly opposed to, the one relying on the mass, automated collection and processing of very large amounts of data.

As the later is becoming increasingly popular, to the extent of being often considered as equivalent to "AI", it is worth considering the question of whether knowledge engineering approaches are still relevant, and of what they can contribute to a broader view of AI and data science. In this presentation, Mathieu d'Aquin will discuss those questions through examples of research illustrating how knowledge engineering approaches play specific roles, for which they are specifically suited, and how those specific roles are becoming increasingly important.