Clustering Sequences of Complex Objects

Sequential Data This paper is about the unsuperviseddiscovery of patterns in sequencesof compositeobjects. A compositeobject may be describedas a sequenceof other, simpler data. In such cases,not only the nature of the componentsis important, but also the order in which thesecomponentsappear, The presentwork studiesthe problem of generalizingsequences of complex objects. A formal definition of generalizedsequencesis given, and an algorithm is derived. Becauseof the excessivecomputationalcomplexity of this algorithm, a heuristic version is described. Thisalgorithmis thenintegratedin a general-purposeclusteringalgorithm. The result is a knowledge discovery system which is able to analyze any structureddatabaseon the baseof a unified, unsupervisedmechanism.

• Type: Conference
• Year: 1997
• Where: KDD
• Authors: A. Ketterlin
• Keywords: Data Mining, Data This Paper, KDD 1997, Presentwork Studiesthe Problem, Simpler Data