06-04-2010, 10:30 PM
[attachment=3069]
ELEMENTARY MACHINE LEARNING FOR WEBMINING
Algorithms that automatically discovers all pages in a website whose location is different from the location where visitors expect to find them and for selecting the set of navigation links to optimize the benefit to the website are proposed.
It is hard to organize a website such that pages are located where visitors expect to find them
APPROACHES
New search mechanism called SmartSeek is introduced
Machine learning concepts Employs genetic algorithm (GA)
System accepts user feedback
FUNCTIONING OF SMARTSEEK
SmartSeek sets up a database about user objective
It extracts information about the common words and phrases
Looks for similar pages (metasearch engine)
Comparison with other pages
Feedback
Further enhancement
Most Relevant
Moderate Relevancy
Irrelevant
Operators AND (&), OR (|),NOT (~) and CONCAT (-).
After n such iteration
THE AI
Artificial Intelligence (AI) is simply a way of thinking intelligently aims at
Natural language processing
Knowledge representation
Automated reasoning
Machine learning
WEB MINING
Web Mining can be said to have three operations of interest:
finding natural grouping of users or pages
finding URLs which tend to be grouped together
finding the order in which URLs tend to be accessed
ALGORITHMIC APPROACH
Finding Expected Locations
Optimizing the set of Navigation Links
OPTIMIZING SET OF NAVIGATION LINKS
1) First Only
2)Optimize Benefit
3)Optimize time
CONCLUSION & FUTURE DIRECTIONS
Algorithms have been proposed
Websites without clear separation of content and index page
Time Threshold to distinguish between two activities
REFERENCES
1.ibm.com
2.almaden.com
3.Mastering Data Mining by Helenz Hulki
4.How internet works by K.Perl
ÂÂ