Top-down and bottom-up approaches to domain knowledge engineering

This document provides a description of the methods for the design and production of the domain knowledge graph (DKG) that is exploited in the HACID project as the backbone for the implementation of hybrid human-artificial collective intelligence. The DKG represents domain knowledge for the two case studies we have considered, namely medical diagnostics and climate change adaptation management (also referred to as climate services). In the following, we provide the background that informs the methodologies used within the project, differentiating between top-down and bottom-up knowledge engineering approaches (Section 2), we discuss the implementation choices for knowledge engineering (Section 3), and we finally present the approach chosen for information extraction with a preliminary evaluation (Section 4).