An Efficient Approach to find Similar Temporal Association Patterns Performing Only Single Database Scan

Traditional frequent pattern mining algorithms use concept of support which is evaluated to a single numeric value. This concept of traditional support value does not hold good to discover similar temporal association patterns for a known reference sequence and threshold value. In temporal sense, support value is sequence of support values and not a single support value. This requires use of a suitable distance function to find the degree of dissimilarity. In this proposed method, we mainly aim at four research objectives. The first objective is to obtain temporal patterns in a single database scan. The second objective is to introduce the concept of negative support sequence to find temporal association patterns. The third objective is to design novel expressions to find the lower bound support sequence and upper bound support sequence by defining two functions called sum and XOR over temporal patterns. The fourth objective is to validate designed expressions. The proposed approach of finding similar temporal patterns is space efficient as it performs only a single database scan and time efficient as it finds temporally similar patterns using only a single database scan.