Usage Aware AverageClicks

1
Usage Aware Average-Clicks

• Kalyan Beemanapalli
• Ramya Rangarajan

2
Overview

• Introduction
• Method/Approach
• Results
• Conclusion and future work
• References
• Questions

3
Motivation

• Most of the algorithms use one type of
  information- Usage or Domain
• Little work has been done to incorporate usage
  and Domain information into one algorithm
• Recent studies have shown that the performance of
  link analysis algorithms like page rank and HITS
  can be enhanced using usage information
• Average Clicks is yet another link analysis
  algorithm which measures distance between two web
  pages.

4
Definitions

• Page Rank
• HITS
• Average Clicks

5
Why Usage Aware Average Clicks

• Usage Aware Page rank, HITS
• Average Clicks A new distance measure on the
  world-wide web
• What is Average Clicks
• Why Average Clicks

6
Our Approach / Method

• Calculate support values of the web pages from
  usage information
• The average-click distance between a page and its
  link is calculated as
• Average-Click
• The shortest path between two web pages is
  defined as
• Where a Damping Factor
• n Average Number of links on a page

7
Example
8
Evaluation Methodology Results

• Testing to be done on the CLA website of
  University of Minnesota
• Calculate Average-Clicks distance and Usage Aware
  Average-Clicks distance between randomly selected
  web pages
• Have the distances validated by an Expert or the
  web administrator of the website
• Still not sure of how to test these results
  against the actual distance as perceived by the
  users of the website
• Results Still under Construction

9
Conclusion and Future Work

• We have defined a new method to measure the
  distance between two web pages which is based on
  the link structure of the website and the usage
  information.
• The goal is to use the usage information
  available from the web logs to remove the bias
  created by using only the link structure.
• The idea is inspired from other approaches where
  usage information has been used to improve the
  popular link analysis algorithms like Page rank
  and HITS
• In the future we would like to investigate what
  weight should be given to link structure
  information and usage information for best
  definition of users intuition of distance
  between pages
• The values of a,n are defined for the world-wide
  web. These values will be different for intranet
  as we expect the web pages to have considerable
  number of external links.

10
References

• Page Rank Algorithm
• Authoritative sources in a hyperlinked
  environment
• Average-Clicks- A new distance measure on the
  world-wide web
• Usage Aware Page Rank
• Modifications of Kleinbergs HITS Algorithm Using
  Matrix Exponentiation and Web Log Records

11
Questions Suggestions