RptGSP: Mining Repetitive Sequential Patterns without Overlapping from Sequence Database
Yongshun Gong, Xiangjun Dong, Xiqing Han, Ruilian Hou · The Open Cybernetics & Systemics Journal · 2015
Repetitive sequential patterns (RSP) mining plays very important roles and has been widely studied in DNA or genome, but there are only a few relevant approaches focusing on mining RSP from sequence database.Taking sequence for example, traditional sequential pattern mining algorithms only consider that appears at one time when calculating the support of , regardless of at least 4 times that appears within this same data sequence.Accordingly, to catch much more interesting sequential patterns, repetitive property needs to be involved during the mining process.However, currently the most relevant RSP methods focus on DNA analysis considering that they cannot be used for recognizing repetitive patterns on events sequences.Therefore, we propose an approach to determine the number of times a sequence repeatedly makes an appearance in a certain data sequence.The support value of a sequence could be more than 100% as this sequence might repeat in one data sequence, therefore we proposed a strategy to ensure the support range of repetitive sequence still within [0,100%].Finally, we proposed an efficient algorithm, called RptGSP, to discover such repetitive sequential patterns based on improving GSP Algorithm.The experimental results reveal that RptGSP can efficiently discover the repetitive patterns.