New Applications of Formal Concept Analysis: A Need for Original Pattern Domains

We survey the results obtained by our research group (joint work with Jérémy Besson and Löıc Cerf, Kim-Ngan T. Nguyen, Marc Plantevit, and Céline Robardet) concerning the design of pattern domains to support knowledge discovery and information retrieval in arbitrary n-ary relations. Our contribution is related to Formal Concept Analysis and its recent developments in direction of, for instance, Triadic Concept Analysis. We focus on a real data mining perspective. It means that we need for both the design of scalable constraint-based mining algorithms and fault-tolerant approaches to support the discovery of relevant patterns from noisy data.