Research on Typical Algorithms in Negative Sequential Pattern Mining

ABSTRACT

Negative sequential pattern (NSP), which contains both non-occurring and occurring items, is sometimes much more informative than positive sequential pattern (PSP) in many applications and intelligent systems. To date, very limited methods are available to mine NSP due to its intrinsic complexities. Furthermore, there is not a unified definition about negative containment, i.e., how a data sequence contains a negative sequence. The researchers who begin to study negative sequential pattern are often confused by these different definitions and are eager to know the differences among the existing methods. So in this paper, we select four typical existing methods, PNSP, Neg-GSP, e-NSP and NSPM, implement their algorithms by JAVA, and compare their definitions, methods, runtimes and the number of NSPs. Examples and experiments on four datasets clearly show their differences.

Keywords: