Performance prediction using Kernel Canonical Correlation Analysis

The paper deals with the problem of anticipating performance parameters for running SPARQL queries. Canonical correlation analysis (CCA) and its kernel variant (KCCA) identify and quantify the associations between two sets of variables. It maximizes the correlation between a linear combination of the variables in one set and a linear combination of the variables in the other set. It measures the strength of association between two sets of variables. The main aspect of this maximization problem is to keep a high dimensional relationship between two sets of variables into few pairs of canonical variables.