Discovering Robust Knowledge from Dynamic Closed-World Data

No. 1: Agents, AI in Art and Entertainment, Knowledge Representation, and Learning

Volume

Proceedings of the AAAI Conference on Artificial Intelligence, 13

Learning Knowledge Bases

Many applications of knowledge discovery require the knowledge to be consistent with data. Examples include discovering rules for query optimization, database integration, decision support, etc. However, databases usually change over time and make machine-discovered knowledge inconsistent with data. Useful knowledge should be robust against database changes so that it is unlikely to become inconsistent after database changes. This paper defines this notion of robustness, describes how to estimate the robustness of Hornclause rules in closed-world databases, and describes how the robustness estimation can be applied in rule discovery systems.