Transcript document

Problem Addressed
• Attempts to prove that Web Crawl is random
& biased image of Web Graph and does not
assert properties of Web Graph
• Understanding the Hyperlink structure has
allowed the design of Algorithms like
PageRank, HITS etc
Approach
• Generally authors use sociological hypothesis
to model the web graphs & compare the
graphs with existing crawls
• This paper attempts to model the web
crawling process using only Zipf’s law but
arrive at results comparable to existing web
graph models.
Criticism
• Uses Zipf’s law to model the crawling process,
but also uses Zipf’s law in evaluation
• Most of the evaluation criteria used are
commonly observed in graphs following Zipf’s
law
• Concludes that Web Crawls are different from
Web Graphs without proving it
Relation to Course
• Closely related to Social Networks as it
discusses Zipf’s law, Small World
Phenomenon, A/H, Page Ranks etc.
• Related to the study of Crawlers