Filomat 2018 Volume 32, Issue 5, Pages: 1875-1885
https://doi.org/10.2298/FIL1805875Q
Full text ( 511 KB)
Cited by
Mining negative sequential patterns from infrequent positive sequences with 2-level multiple minimum supports
Qiu Ping
Zhao Long
Chen Weiyang
Xu Tiantian
Dong Xiangjun
Negative sequential patterns (NSP) referring to both occurring items
(positive items) and nonoccurring items (negative items) play a very
important role in many real applications. Very few methods have been
proposed to mine NSP and most of them only mine NSP from frequent positive
sequences, not from infrequent positive sequences (IPS). In fact, many
useful NSP can be mined from IPS, just like many useful negative association
rules can be obtained from infrequent itemsets. e-NSPFI is a method to mine
NSP from IPS, but its constraint is very strict to IPS and many useful NSP
would be missed. In addition, e-NSPFI only uses a single minimum support,
which implicitly assumes that all items in the database are of the similar
frequencies. In order to solve the above problems and optimize NSP mining, a
2-level multiple minimum supports (2-LMMS) constraint to IPS is proposed in
this paper. Firstly, we design two minimum supports constraints to mine
frequent and infrequent positive sequences. Secondly, we use Select
Actionable Pattern (SAP) method to select actionable NSP. Finally, we
propose a corresponding algorithm msNSPFI to mine actionable NSP from IPS
with 2-LMMS. Experiment results show that msNSPFI is very efficient for mining
actionable NSP.
Keywords: Negative sequential patterns, infrequent positive sequences, multiple minimum supports, actionable