This article presents the use of Answer Set Programming (ASP) to minesequential patterns. ASP is a high-level declarative logic programming paradigmfor high level encoding combinatorial and optimization problem solving as wellas knowledge representation and reasoning. Thus, ASP is a good candidate forimplementing pattern mining with background knowledge, which has been a datamining issue for a long time. We propose encodings of the classical sequentialpattern mining tasks within two representations of embeddings (fill-gaps vsskip-gaps) and for various kinds of patterns: frequent, constrained andcondensed. We compare the computational performance of these encodings witheach other to get a good insight into the efficiency of ASP encodings. Theresults show that the fill-gaps strategy is better on real problems due tolower memory consumption. Finally, compared to a constraint programmingapproach (CPSM), another declarative programming paradigm, our proposal showedcomparable performance.
展开▼