An enhanced transaction identification module on Web usage mining

AsiαPac!fìc Mallagemenr Rev;ew

(2001) 6(2). 241-252

An enhanced transaction identification module on Web usage mining

C. c. Hs ieh'and C.T. Chang

Web data mining is a process which transforms unor ganized or le ss-structured data on the World Widc W eb into usc ful infom131Îo

Web usage

one kind of W eb data mining f OCllses 011 exploring users' browsing pattem on a single Wcb site from Web acccss data restored in Ihe Web server. Web usage mining in

0

Ihree S I CpS: (1) clean!filler Ihe Web access data. (2) convert thc W eb access data into a set of transaction s by deploying transaction identification modules. and (3) explore the relat i onship of interest in thc co nverted transactions

lI

sing data mining techniques. The quality ofthe rclationship being discovered fr0111 thc transac- lion set large ly

the data convcrsion

might be lost during data cO

version

3nd cffectivc transaction identification modules arc thus necessary. In this study

wc proposc an enha

ced transaction identificati on module. The simulation rcsults show thc cff cctiveness 0

our proposed m ethod comparcd with th e earl y studies

KeY^lI'ordr

Data Mining: Association

Identification Module: W cb Usage Mining

\,

Introduction

As lllore and lllore bus iness is conducted electronically , for insta nce , usin g the Internet, the custolller data acculllu late s daily in an ever increasing quanti ty. From marketing viewpoint , being able to explore useful informa- tion from th e enormous alllount of data to aid marketing strategies and deci- sion makin g is of

mining is such a technology for data analysis that, depending upon the application , generates one or more types of informalion : (1) association , (2)

(3) classification, and (4)

The literature in data mining can be referred to in [3] and the ref- erences therein

Web data lllining is to discover useful infomnation on the World Wide Web (WWW)

sing data mining technology. Two categories of Web data mining are present: (1) Web content

which explores useful information on the WWW [2, 7 , 10], and (2) Web usage mining, which mines users' browsing behavior on a Web site [5 , 6, 8, 9]. In this paper, Web usage mining is assumed

A Web page might serve as either a navigation page

by which users

. Corresponding author.

partmcnt oflndustrÎal Manage

ent Science. National Cheng Kung Universily. Tainan 70

Taiwan. R.O.

FAX : +886-6-236-2162; E-mail

[email protected]

241