High performance computing in database mining

Describes how parallel processing can be used to speed up database mining operations. One of the most common database mining tasks is that of inducing relationships which hold between attributes of a database relation. These relationships may take the form of IF-THEN rules. Rules which are discovered to apply to all of the extensional database are called strong rules. Strong rules are useful for decision support and are often used in expert systems and expert database systems. The algorithm we describe, STRIP (Strong Rule Induction In Parallel), discovers strong rules within a database relation, using parallel techniques. We have shown how the rules induced using STRIP can be used for implementing a semantic query pre-processor, among other potential uses