Skip to main navigation Skip to search Skip to main content

Nonparametric analysis of extremes on web graphs: PageRank versus max-linear model

  • Natalia M. Markovich
  • , Maxim Ryzhov
  • , Udo R. Krieger

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

9 Scopus citations

Abstract

We analyze the cluster structure in large networks by means of clusters of exceedances regarding the influence characteristics of nodes. As the latter characteristics we use PageRank and the Max-Linear model and compare their distributions and dependence structure. Due to the heaviness of tail and dependence of PageRank and Max-Linear model observations, the influence indices appear by clusters or conglomerates of nodes grouped around influential nodes. The mean size of such clusters is determined by a so called extremal index. It is related to the tail index that indicates the heaviness of the distribution tail. We consider graphs of Web pages and partition them into clusters of nodes by their influence.

Original languageEnglish
Title of host publicationDistributed Computer and Communication Networks - 20th International Conference, DCCN 2017, Proceedings
EditorsKonstantin E. Samouylov, Vladimir M. Vishnevskiy, Dmitry V. Kozyrev
PublisherSpringer Verlag
Pages13-26
Number of pages14
ISBN (Print)9783319668352
DOIs
StatePublished - 1 Jan 2017
Externally publishedYes
Event20th International Conference on Distributed Computer and Communication Networks, DCCN 2017 - Moscow, Russian Federation
Duration: 25 Sep 201729 Sep 2017

Publication series

NameCommunications in Computer and Information Science
Volume700
ISSN (Print)1865-0929

Conference

Conference20th International Conference on Distributed Computer and Communication Networks, DCCN 2017
Country/TerritoryRussian Federation
CityMoscow
Period25/09/1729/09/17

Keywords

  • Extremal index
  • Max-Linear model
  • PageRank
  • Tail index
  • Web graph

ASJC Scopus subject areas

  • General Computer Science
  • General Mathematics

Fingerprint

Dive into the research topics of 'Nonparametric analysis of extremes on web graphs: PageRank versus max-linear model'. Together they form a unique fingerprint.

Cite this