Uncertain Frequent Pattern Mining - Frequent Pattern Mining - page 334

Chapter 14
Uncertain Frequent Pattern Mining

Carson Kai-Sang Leung

Abstract Frequent pattern mining aims to discover implicit, previously unknown and
potentially useful knowledge—in the form of frequently occurring sets of items—that
are embedded in data. Many of the models and algorithms developed in the early days
mine frequent patterns from traditional transaction databases of precise data such as
shopper market basket data, in which the contents of databases are known. However,
we are living in an uncertain world, in which uncertain data can be found in various
real-life applications. Hence, in recent years, researchers have paid more attention
to frequent pattern mining from probabilistic datasets of uncertain data. This chapter
covers key models, algorithms and topics about uncertain frequent pattern mining.

Keywords Data mining · Knowledge discovery from uncertain data · Associa-

tion rule mining · Frequent patterns · Frequent itemsets · Probabilistic approach ·

Uncertain data

1 Introduction

As an important data mining task, frequent pattern mining [ 8 , 12 ] aims to discover
implicit, previously unknown and potentially useful knowledge—revealing patterns
on collections of frequently co-occurring items, objects or events—that are em-
bedded in data. Nowadays, frequent pattern mining is commonly used in various
real-life business, government, and science applications (e.g., banking, bioinformat-
ics, environmental modeling, epidemiology, finance, marketing, medical diagnosis,
meteorological data analysis). Uncertain data are present in many of these appli-
cations. Uncertainty can be caused by (i) our limited perception or understanding
of reality; (ii) limitations of the observation equipment; or (iii) limitations of avail-
able resources for the collection, storage, transformation, or analysis of data. It
can also be inherent in nature (say, due to prejudice). Data collected by acoustic,
chemical, electromagnetic, mechanical, optical radiation, thermal sensors [ 6 ]inen-
vironment surveillance, security, and manufacturing systems can be noisy. Dynamic