Akademik Veri Yönetim Sistemi
3rd International Conference on Data Warehousing and Knowledge Discovery, DaWaK 2001, Munich, Almanya, 5 - 07 Eylül 2001, cilt.2114, ss.83-92
© Springer-Verlag Berlin Heidelberg 2001.In sequential pattern mining, the support of the sequential pattern for the transaction database is defined only by the fraction of the customers supporting this sequence, which is known as the customer support. In this paper, a new parameter is introduced for each customer, called as repetition support, as an additional constraint to specify the minimum number of repetitions of the patterns by each customer. We call the patterns discovered using this technique as cyclically repeated patterns. The additional parameter makes the new mining technique more efficient and also helps discovering more useful patterns by reducing the number of patterns searched. Also, ordinary sequential pattern mining can be represented as a special case of the cyclically repeated pattern mining. In this paper, we introduce the concept of mining cyclically repeated patterns, we describe the related algorithms, and at the end of the paper we give some performance results.