Ontologies and Description Logics

Ontologies and Description Logics (DLs). Why putting them in a single chapter for this document devoted to AI research in Italy? Well, a good reason is that both research areas originated from a single, classic mainstream in AI: logic-based knowledge representation (KR). Indeed, starting from the late 80s, Italian researchers brought a radical contribution to the formal understanding of so-called structured representation languages developed in the previous decade, whose most known example was KL-ONE. Their research took however two different directions. On one hand, after Levesque and Brachman’s logicalization of KL-ONE and their discovery of the fundamental tradeoff between expressivity and tractability [8], a whole Italian school emerged under the direction of Maurizio Lenzerini, focusing mainly on the need to develop KR languages with predictable computational behavior and sound and complete inference algorithms. On the other hand, Nicola Guarino started analyzing the implicit and often ambiguous ontological assumptions made by structured logical formalisms, such as those behind the notions of role and attribute, arguing against the ontological neutrality of representation systems and suggesting formal ontological distinctions to be taken into account. In the meanwhile, in the early 90s, a new hot topic captured the attention of AI researchers: knowledge sharing among heterogeneous, distributed sources (very similar to the issue emerging –roughly at the same time– in the DB community: data integration). It became clear soon that, in order to attack these problems, two complementary aspects needed to be addressed: content and reasoning. Ontological analysis and DLs emerged soon as the proper techniques to deal with these aspects, and became later very popular for their key role in the future Semantic Web. Italian researchers contributed in a peculiar way to both.