ECML/PKDD-2002 Tutorial

Web Usage Mining for E-business Applications

19 August 2002 14:30-18:00

Index

Motivation ·
Objectives ·
Target Audience ·
Outline ·
Presenters ·
Related Links ·

Motivation

Web usage mining has become a critical tool for competitive business intelligence. Understanding the behavior of a site's visitors requires creative applications of KDD techniques to e-commerce and clickstream data: patterns must be discovered from a variety of data sources, and these patterns must be interpreted and transformed into actionable knowledge for redesigns that bring revenue. Redesigns encompass general improvements to information architecture and navigation options, as well as the offering of personalized recommendations and services.

Objectives

In this tutorial, we go in depth into the KDD challenges in Web usage mining. In particular, we address

i) the reliability of the data in the presence of noise, data anonymization and deliberate data corruption;

ii) methodologies to derive interesting patterns from the data and measures of pattern interestingness, importance or actionability; and

iii) techniques for exploiting KDD results and methods for evaluating their impact.

The tutorial draws from the core domains of KDD, in which issues on data preparation and methodological analysis are covered. We also draw on the domain of web marketing that contributes the requirements and the economic measures, and from the domain of human-computer interaction that supplies the evaluation methodologies.

Target Audience

ECML/PKDD participants with interest in e-applications in general and in Web usage mining in particular. This group includes

i) ML and KDD researchers who are interested in gaining insight to the challenges, solutions and open issues of this exciting domain;

ii) e-commerce practitioners who are already familiar with KDD basics and want to learn more about KDD enabling technologies for measuring, evaluating, and improving e-services.

Relevant topics of the ECML/PKDD conference

business problems and their data mining solutions ·
data preparation and integration ·
quality assessment ·
use of prior knowledge ·
applications: marketing, personalization, information systems ·

Outline

Part I. Introduction

Motivation and background: Economic and HCI perspectives ·
Overview ·

Part II. Data acquisition and data preparation for Web usage mining

Obtrusive and non-obtrusive data collection ·
Types and sources of data ·
Data preparation ·
Integrating content and structure data ·

Part III. The goals: Metrics for site success

Business-oriented metrics: conversion rates, E-Metrics, etc. ·
User-oriented metrics: user satisfaction, other usability measures ·

Part IV. The techniques: enabling technologies

Pattern discovery techniques: General ·
Specific pattern discovery techniques in Web usage mining ·
Evaluation of discovered patterns / The Post-Mining Phase ·

Part V. Applications and challenges

Customer conversion and site success evaluation: a case study ·
Personalization ·
Privacy Concerns ·
Research issues / future directions ·

Presenters

Myra Spiliopoulou is professor of e-business at the Leipzig Graduate School of Economics. Her current research includes web usage mining, mining of implicitly structured texts, KDD for web-merchandizing, pattern maintenance and pattern evolution. Her teaching curriculum includes courses on web mining and e-business. She is co-chair of the series of the KDD Conference workshops on web mining WEBKDD'99, WEBKDD'2000, WEBKDD'01 and WEBKDD'02. At the previous PKDDs, she presented tutorials on web mining, with emphasis on KDD methodologies (PKDD'99) and on evaluation methods (PKDD'2000). Together with Bamshad Mobasher and Bettina Berendt, she has presented a tutorial on Web mining with emphasis on personalization at ECML/PKDD'2001.

Bamshad Mobasher is an assistant professor of Computer Science and the director of the Center for Web Intelligence at DePaul University. His current research includes the applications of Web usage mining for automatic personalization, multi-agent systems for automated contracting, and intelligent Web agents involving the applications of text mining and machine learning, also offering courses on the above subjects. Dr. Mobasher is further overseeing several joint projects with the industry in these areas. He was a co-organizer the Workshop on Intelligent Techniques for Web Personalization, held at IJCAI 2001, Seattle, Washington, in August 2001. Together with Bettina Berendt and Myra Spiliopoulou, he has presented a tutorial on Web mining with emphasis on personalization at ECML/PKDD'2001.

Bettina Berendt is a Scientific Assistant at the Institute of Information Systems at Humboldt University Berlin. Her research interests include web usage mining, psychological methods of web navigation analysis, and visualization. Until last year, Dr. Berendt was the director of "SchulWeb" (http://www.schulweb.de/), a large non-commercial German web server. Bettina Berendt's teaching experience includes seminars and tutorials on Web mining, AI and Cognitive Science, and Visualization on the Web. She is a co-organizer of the ECML/PKDD workshops Semantic Web Mining 2001 and Semantic Web Mining 2002. Together with Bamshad Mobasher and Myra Spiliopoulou, she has presented a tutorial on Web mining with emphasis on personalization at ECML/PKDD'2001.