Knowledge bases such as Wikidata, YAGO, or the Google Knowledge Graph are increasingly used in applications like entity resolution, structured search or question answering. The correctness of these knowledge bases is generally well understood and at the center of attention in automated construction techniques. About their completeness, in contrast, most knowledge is anecdotal: Completeness is usually correlated with popularity of topics, yet understanding whether a knowledge base has complete information on a specific topic is often difficult.
In this talk I will introduce the problem of understanding the completeness of knowledge bases. I will present three approaches: (i) data-inherent inferences about completeness using association rule mining, (ii) extraction of cardinality information from text sources, and (iii) various tools we developed to help knowledge base curators and consumers in mapping and understanding the completeness of knowledge bases.