cSPADE -UE: Algorithm for Sequence Mining for Unstructured Elements Using Time Gap Constraints

-We present a new state machine that combines two techniques for complex data sequences: Data modeling and frequent sequence mining. This algorithm relies on unstructured variable gap sequence miner, to mine frequent patterns with different gap between elements. Here we will have two variations: Sequence pruning technique for other primary frequent sequences to reduce space complexity and allow creating same sequence to form even if they do not have matching on all the positions. We apply algorithm to task of protein sequence classification on real data from protein families. A state of the art method for protein classification, by decreasing the state space complexity and improving the accuracy.