Network Modeling of Consumers' Selection of Providers Based on Online Reviews

Tian Gan, Rishita Das, Maurizio Porfiri

Research output: Contribution to journalArticlepeer-review


Information spreading over online review systems affects people's opinions and choices. In this work, we study consumers' decision-making process with respect to provider selection, accounting for the providers' online reviews and accessibility. We propose a network-based dynamical system, in which the consumers switch between providers based on online reviews, as the time-varying online review system is continuously updated by the consumer fluxes. We apply the model to various network structures, capturing providers' accessibility: i) random, canonical networks and ii) a real-world network of medical doctors in New York City. We examine the emerging correlations and causal relationships between the success of providers and the topological properties of the networks. Across a wide range of networks of varying size, we consistently find that online reviews have an important role in providers' success. The satisfaction of the consumers in the online review systems, together with the market share, influences consumer fluxes between providers and the overall quality of service experienced by consumers. The study of the network of doctors reveals some causal mechanisms in the decision-making processes, with the doctor's success impacting on the providers' quality of service and the consumer fluxes.

Original languageEnglish (US)
Pages (from-to)2757-2768
Number of pages12
JournalIEEE Transactions on Network Science and Engineering
Issue number3
StatePublished - May 1 2024


  • Complex systems
  • Markov chain
  • decision-making
  • online reviews
  • urban data

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Computer Networks and Communications
  • Computer Science Applications


Dive into the research topics of 'Network Modeling of Consumers' Selection of Providers Based on Online Reviews'. Together they form a unique fingerprint.

Cite this