張忠義老師的歷程檔案

論文名稱	20090814--Sequential Patterns Mining with Fuzzy Time-Intervals
發表日期	2009-08-14

[英文摘要] :
The task of sequential pattern mining is useful for various
applications, including market analysis, decision support,
and business management. One important issue is to
discover frequent sequential patterns in a sequence
database. And most of the previous works have focus on
the order of times. However, the time interval between
successive items in patterns is seldom discussed before.
With the order of items, sequential pattern is not as good
as which is extended with time interval to make the
decision. In this paper, we propose an algorithm called
sequential pattern mining with fuzzy time intervals
(SPFTI). The main idea of SPFTI algorithm is to use the
Apriori-like method to mine the frequent sequential
patterns of sequence database and use fuzzy theory to
mine the time interval between frequent sequences. At
first, find the candidate sequential patterns. Then, the
frequent sequential patterns are found with the minimum
support. In the step of finding frequent sequential
patterns, use the fuzzy number to find each time cluster by
computing its fuzzy support. And the results are the
frequent fuzzy time sequential patterns. Finally, the
experimental result verifies that result of our proposed
SPFTI algorithm outperforms with the fuzzy sequential
patterns mining with fixed time interval