System and method for defining shapes with which to mine time sequences in computerized databases

摘要

A system and method including a computer shape definition language are disclosed for defining shapes and mining time sequences that resemble the shapes. The system and method include provisions for establishing a user- defined alphabet that in turn establishes a set of elemental shapes. The system also includes simple yet powerful operators for combining the elemental shapes to define a desired time sequence shape. Moreover, intervals of actual time sequences are mapped into corresponding transition sequences using the alphabet, and the transition sequences are stored in a hierarchical index structure for easily accessing the transition sequences. The index structure is entered with the desired time sequence shape, and the index structure is traversed to identify maximal actual transition sequences which conform to the desired time sequence shape, within user-definable blurry criteria.