Description Logics (DL) are a successful family of logic-based knowledge
representation languages, which can be used to represent the conceptual
knowledge of an application domain in a structured and formally
well-understood way. They are employed in various application domains,
such as natural language processing, databases, the semantic web, and
biomedical ontologies. As the size of DL knowledge bases grows, tools
that support improving their quality become more important. Standard DL
reasoning can be used to computed implicit consequences such as
inconsistencies and inferred subsumption relationships, but it does not
explain the reasons for a given consequence.

Axiom pinpointing is a first step towards providing such an explanation.
Given a DL knowledge base and a consequence, it computes minimal
(maximal) subsets of the KB that have the consequence (do not have the
consequence). In the talk, I will review recent results on pinpointing 
in the DL $\mathcal{EL}$. Though of limited expressive power, 
$\mathcal{EL}$ is 
used in large biomedical ontologies such as Snomed and the Gene Ontology.