Use of supervised learning to predict directionality of links in a network

Sucheta Soundarajan, John E. Hopcroft

Research output: Chapter in Book/Entry/PoemConference contribution

4 Scopus citations

Abstract

Often, the information contained in network data is incomplete. Many avenues of research are aimed at addressing this incompleteness. For example, the link prediction problem attempts to identify which missing links are most likely to exist in the complete network. In this paper, we consider a related, but different, problem: predicting the directions of links in a directed network. We treat this problem as a supervised learning problem in which the directions of some edges are known. We calculate various features of each known edge based on its position in the network, and use a Support Vector Machine to predict the unknown directions of edges. We consider four networks, and show that in each case, this method performs significantly better than other compared methods.

Original languageEnglish (US)
Title of host publicationAdvanced Data Mining and Applications - 8th International Conference, ADMA 2012, Proceedings
Pages395-406
Number of pages12
DOIs
StatePublished - 2012
Externally publishedYes
Event8th International Conference on Advanced Data Mining and Applications, ADMA 2012 - Nanjing, China
Duration: Dec 15 2012Dec 18 2012

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7713 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other8th International Conference on Advanced Data Mining and Applications, ADMA 2012
Country/TerritoryChina
CityNanjing
Period12/15/1212/18/12

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Use of supervised learning to predict directionality of links in a network'. Together they form a unique fingerprint.

Cite this