Ontology Design Patterns in Ontology Learning

Ontology Learning (OL), either from text or other resources, suffers from problems related to implicit knowledge being incompletely extracted, knowledge that is too specific for creating a general domain model, and data or text structures that are not easily transformed into a high-quality ontology. Many types of background knowledge can be used to remedy problems of implicit or missing knowledge, however applying an appropriate modeling solution and style is a harder problem. Ontology Design Patterns (ODPs) are encodings of general best practices in ontology design. There are many kinds of patterns that can contribute to OL systems in different ways. In this chapter we first present the kinds of ODPs that exist and the kinds of problems they address, we then discuss how ODPs could potentially contribute to OL. Finally, we take two examples of research approaches where ODPs have been used for improving OL results or perform OL in novel ways. An overall conclusion is that ODPs can both provide guidance for OL, in terms of task-focus and making design choices available in a formalized manner, as well as being reusable building blocks that could potentially raise the focus of OL from single elements to detection and composition of larger complex components.