Adaptive XML Tree Classification on Evolving Data Streams
We propose a new method to classify patterns, using closed and maximal frequent patterns as features. Generally, classification requires a previous mapping from the patterns to classify to vectors of features, and frequent patterns have been used as features in the past. Closed patterns maintain the same information as frequent patterns using less space and maximal patterns maintain approximate information. We use them to reduce the number of classification features. We present a new framework for XML tree stream classification. For the first component of our classification framework, we use closed tree mining algorithms for evolving data streams. For the second component, we use state of the art classification methods for data streams. To the best of our knowledge this is the first work on tree classification in streaming data varying with time. We give a first experimental evaluation of the proposed classification method.