Gauging node consistency in accusation-endorsement networks

Oscar Goodloe, Zihan Zhou, Joel Nishimura

Research output: Contribution to journalArticlepeer-review


Many signed, directed social networks can be viewed as being composed of positive (endorsements) and negative (accusations) directed edges, and these networks can in turn be created through a variety of different processes. The recently proposed consistency dynamics supposes that when nodes expect to be judged based on their associations in the network, they may create edges out of a desire to appear as having consistent judgements. We develop a quantifiable score that can rate the level of consistency in a node's judgement. We demonstrate that this consistency score can be efficiently estimated using a modification of the popular personalized PageRank algorithm and evaluate the score's properties. In order to validate this score's relevance to empirical networks, we use consistency scores to perform an edge prediction task, and demonstrate that it performs competitively with, and adds complementary information to, more complicated measures designed specifically for that task. We also demonstrate that the nodes in these networks exhibit specific behaviours that consistency can identify across a range of parameterization values and which are not recoverable by other measures in isolation.

Original languageEnglish (US)
Article number3
JournalJournal of Complex Networks
Issue number3
StatePublished - Jun 1 2022


  • PageRank
  • link sign prediction
  • network motif
  • signed network

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Management Science and Operations Research
  • Control and Optimization
  • Computational Mathematics
  • Applied Mathematics


Dive into the research topics of 'Gauging node consistency in accusation-endorsement networks'. Together they form a unique fingerprint.

Cite this