TY - JOUR
T1 - On mining progressive positive and negative sequential patterns simultaneously
AU - Huang, Jen Wei
AU - Wu, Yong Bin
AU - Jaysawal, Bijay Prasad
N1 - Publisher Copyright:
© 2020 Institute of Information Science. All rights reserved.
PY - 2020
Y1 - 2020
N2 - Positive sequential pattern (PSP) mining focuses on appearing items, while negative sequential pattern (NSP) mining tends to find the relationship between occurring and non-occurring items. There are few works involved in NSP mining, and the definitions of NSP are inconsistent in each work. The support threshold for PSP is always applied on NSP, which cannot bring out interesting patterns. In addition, PSP has been discovered on incremental databases and progressive databases, while NSP mining is only performed on static databases. Progressive sequential pattern mining finds the most up-to-date patterns, which can provide more valuable information. However, the previous progressive sequential pattern mining algorithm contains some redundant process. In this paper, we aim to find NSP on progressive databases. A new definition of NSP is given to discover more meaningful and interesting patterns. We propose an algorithm, Propone, for efficient mining process. We also propose a level-order traversal strategy and a pruning strategy to reduce the calculation time and the number of negative sequential candidates (NSC). By comparing Propone with some modified previous algorithms, the experimental results show that Propone outperforms comparative algorithms.
AB - Positive sequential pattern (PSP) mining focuses on appearing items, while negative sequential pattern (NSP) mining tends to find the relationship between occurring and non-occurring items. There are few works involved in NSP mining, and the definitions of NSP are inconsistent in each work. The support threshold for PSP is always applied on NSP, which cannot bring out interesting patterns. In addition, PSP has been discovered on incremental databases and progressive databases, while NSP mining is only performed on static databases. Progressive sequential pattern mining finds the most up-to-date patterns, which can provide more valuable information. However, the previous progressive sequential pattern mining algorithm contains some redundant process. In this paper, we aim to find NSP on progressive databases. A new definition of NSP is given to discover more meaningful and interesting patterns. We propose an algorithm, Propone, for efficient mining process. We also propose a level-order traversal strategy and a pruning strategy to reduce the calculation time and the number of negative sequential candidates (NSC). By comparing Propone with some modified previous algorithms, the experimental results show that Propone outperforms comparative algorithms.
UR - http://www.scopus.com/inward/record.url?scp=85078941781&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85078941781&partnerID=8YFLogxK
U2 - 10.6688/JISE.20200136(1).0009
DO - 10.6688/JISE.20200136(1).0009
M3 - Article
AN - SCOPUS:85078941781
SN - 1016-2364
VL - 36
SP - 145
EP - 169
JO - Journal of Information Science and Engineering
JF - Journal of Information Science and Engineering
IS - 1
ER -