Web Log Analyzer

1
Web Log Analyzer

• Term Project Report - CSCI 8710
• Zhibin Huang
• Computer Science Department, UGA

2
Log file

• Log file is a file that lists actions that have
  occurred
• Log file analysis is important to the website

3
Our Goal

• Given two log files
• trichdb.log from http//www.trichdb.org/trichdb/
• toxodb.log from http//toxodb.org/toxo/
• data clustering
• characterize user clusters
• Simplified Customer Behavior Model Graph (CBMG)

4
Log file format

• user_ip, time, request object, status code,
  transfer volume, request url, client_info
• 152.77.24.38 - - 10/Nov/2008080350 -0500
  "GET /gbrowse/tmp/toxodb/img/a599e46b342f1a4bb389b
  d8d9a6d57b3.png HTTP/1.0" 200 1575
  "http//www.toxodb.org/cgi-bin/gbrowse/toxodb/"
  "Mozilla/5.0 (Windows U Windows NT 5.1 fr
  rv1.9.0.3) Gecko/2008092417 Firefox/3.0.3"

5
Data analysis

• Data Preprocessing
• Extract unique user_ip
• Data clustering
• user_ip -gt user region
• based on the web service of website
  http//www.ip2location.com/

6
User region analysis
7
Analysis of user time
8
Friendly usage
9
CBMG model for trichdb
10
Matrix of Transition Probabilities for trichdb

11
Data analysis of toxodb
12
Data analysis of toxodb(cont)
13
Data analysis of toxodb(cont)
14
Matrix of Transition Probabilities toxodb
15
Conclusion

• Users in U.S. are more efficient in using these
  two websites than people in other countries.
• Website toxodb is much easier for user to use
  than website trichdb.

16
QA

• Thanks