Pattern Discovery from Usage, Content, and Structure

10/19/00


Click here to start


Table of Contents

Pattern Discovery from Usage, Content, and Structure

Web Usage Mining Process

Pattern Discovery

Common Data Mining Tasks

Sequential Patterns

Finding Frequent Navigational Patterns

Finding Frequent Navigational Patterns

Filtering Patterns Based on “Interestingness”

Finding Interesting Patterns

Example: Algorithms for Filtering Frequent Itemset

Discovering Aggregate Usage Profiles

Methodologies for the Discovery of Aggregate Profiles

Profile Aggregation Based on Clustering Transactions (PACT)

PACT - An Example

Hypergraph-Based Clustering

Attaching Weights to Hyperedges

Selection of Good Partitions

Strengths of ARHP

Usage Profiles Based on Hypergraph Clusters of Pageviews

Aggregate Usage Profiles - Hypergraph

Aggregate Usage Profiles - PACT

Integrating Content and Usage

Discovery of Content Profiles

How Content Profiles Were Generated

How Content Profiles Were Generated

How Content Profiles Were Generated

User Segments Based on Content

Another Example of Web Content Mining: WebACE (Mobasher, Boley, Gini, Han - 1998)

WebACE Architecture

What are the Transactions?

Experiments with WebACE

Feature Selection Criteria Used in the Experiments

Entropy Comparison of Algorithms

Document Clusters in AutoClass

Document Clusters in ARHP

Applying ARHP to S&P Stock Data

Clustering of S&P 500 Stock Data

Author: Bamshad Mobasher

Email: mobasher@cs.depaul.edu

Home Page: http://maya.cs.depaul.edu/~mobasher/classes/cs589

Download presentation source