An OLAM Approach to Analyze

1
1º Simpósio Doutoral do Departamento de
Informática
Braga, Portugal, 16, outubro, 2003
An OLAM Approach to Analyze e-commerce
Clickstreams
Ronnie Alves, Orlando Belo ronnie,
obelo_at_di.uminho.pt
Departamento de Informática Escola de
Engenharia Universidade do Minho PORTUGAL
2
Abstract

• Clickstream analysis can reveal usage patterns on
  the companys web site and give highly improved
  understanding of customer behaviour. This
  understanding can then be utilized for improving
  customer satisfaction with the web site and the
  company in general, yelding great business
  advantage. Such summary information and rules
  have to be extracted from very large collections
  of clickstreams in web sites. This is challenging
  data mining, both in terms of the magnitude of
  data involved, and the need to incrementally
  adapt the mined patterns and rules as new data is
  collected. In this paper, we present some
  guidelines for implementing on-line analytical
  mining (OLAM) engines. In addition, we describe
  our e-Miner project called CubeMiner, which is an
  OLAM approach for analyzing clickstreams.
  Besides, we based our research on implementations
  that we consider efficient approaches on on-line
  analytical mining systems, such as DBMiner and
  WebLobMiner from Simon Fraser University (CA),
  and the OLAP-based Web Access Engine from HP Labs
  (USA). Moreover, our previous experience in
  clickstream analysis using datasets from KDD Cup
  2000 (focus on web mining tasks) is also
  discussed.

3
Contents

• Motivation
• On-Line Analytical Mining (OLAM)
• OLAM Research
• Our OLAM project CUBEMINER
• Experience on Clickstream Analysis
• Conclusions and Future Work

4
Motivation

• Decision Support Crucial
• Means to analyze large volumes of data
  clickstreams
• Improving customer satisfaction, provides
  competitive advantage
• More than 40 commercially available applications
  for clickstream analysis
• Only statistical reports, and diferent
  statistical results
• New tools has appeared
• Web Usage Mining tools
• The recent progress of data mining and data
  warehouse Fayyad et al 1998
• Integrating OLAP with data mining

5
OLAM

• On-Line Analytical Mining (OLAM)
• Is a mechanism which integrates on-line
  analytical processing (OLAP) with data mining
• Desired mining functions of OLAM engines
• Data Characterization, Class Comparison,
  Association, Prediction, Classification,
  Time-series Analysis, and so on
• Including OLAP operations
• An exploratory process in nature
• Mining should be performed at different portions
  of data at multiple levels of abstraction.

6
OLAM Research

• DBMiner, from Simon Fraser University
• The general architecture of DBMiner, tightly
  integrates a relational database system, with a
  concept hierarchy module, and a set of knowledge
  discovery modules
• WebLogMiner, from Simon Fraser University
• Data Mining techniques are put to use with web
  log data cube to predict, classify, and
  discoverer interesting correlations
• Special attention has been taken in time-series
  analysis
• An OLAP based Web Access Engine, from HP Labs
• It is possible to overcome the performance
  problems of handling sparce data cubes
• They use the OLAP server as computing engine to
  support data mining operation

7
CUBEMINER

• The main goal of the CUBEMINER Project is to
  develop an OLAM engine based on Data Mining and
  OLAP techniques to analyze specialized
  clickstrems from specialized data cubes.

8
Experience on Clickstream Analysis

• Data Understanding and Preparation
• Question 1 to predict when an user leave or
  continues its navigation on the web site
• KDD Cup 2000 Dataset.
• Data Modeling
• Decision tree technique for predicting session
  continues
• WEKA system, Association Mining (J48) C4.5
  revision 8
• Results
• Sessions with length lt 1
• Sessions with length gt 1

9
Final Remarks

• Currently available web log analysis tools
  provide only statistical reports, and they are
  generally restricted to the datas size
• Integrating OLAP with Data Mining techniques
  should be an alternative to make the clickstreams
  analysis more discovery driven.
• Some guidelines on OLAM Projects has been
  presented

10
Current Research and Development Work

• Crawlers Detection and Filtering plaform

11
Current Research and Development Work

• CubeMiners Engine Architecture
• Cube Definition
• Cube Querying
• Cube Mining

12
1º Simpósio Doutoral do Departamento de
Informática
Braga, Portugal, 16, outubro, 2003
An OLAM Approach to Analyze e-commerce
Clickstreams
Ronnie Alves, Orlando Belo ronnie,
obelo_at_di.uminho.pt
Departamento de Informática Escola de
Engenharia Universidade do Minho PORTUGAL