Published online by Cambridge University Press: 31 January 2011
Data mining of crystal structure databases is performed with methods using deduction, semiempirical heuristic principles, or artificial intelligence to create new information from crystal structure data. In this article, enlightening examples, ranging from a systematic search for new ferroelectric materials to ultrahard, ductile, or high-dielectric-permittivity materials, are detailed, with numerous references providing a handle on the state of the art. This activity is just leaving infancy, especially for its ab initio aspects, with the brightest future ahead. Data mining will clearly guide expensive experiments through inexpensive computations for decades to come. This perspective greatly enhances the value of structure data patiently archived within crystal structure databases for nearly a century now.