EventMiner Framework

Abstract

pattern formulation and pattern mining language. The language is composed of a well-defined set of operators that facilitate pattern analysis. Hypothesis forma­ tion is achieved by data-driven operators that bring hidden interesting patterns to the surface. Hypothesis testing is accomplished by hypothesis-driven operators that facilitate knowledge formulation and pattern query. Data-driven operators are used to generate a basic model and derive a prelim­ inary insight. Then, an analyst can seed a hypothesis and grow it step by step using hypothesis-driven operators. A good hypothesis is one that is not necessarily correct but one that opens new paths of investigation. This path cannot be fully perceived in advance in complex modeling tasks. So the analyst must be provided with appropriate operators to carry out new experiments based on the original hypothesis. EventMiner Framework