TrustFlow: Topic-Aware Vector Reputation Propagation for Multi-Agent Ecosystems

TL;DR

TrustFlow is a novel reputation propagation algorithm that assigns multi-dimensional reputation vectors to agents, leveraging topic-aware transfer operators for robustness and direct queryability in multi-agent systems.

Contribution

It introduces a topic-aware, vector-based reputation propagation method with provable convergence and robustness against attacks, outperforming traditional scalar reputation models.

Findings

Achieves up to 98% Precision@5 on dense graphs

Resists sybil attacks and reputation laundering with minimal impact

Produces directly queryable vector reputations in the same embedding space

Abstract

We introduce TrustFlow, a reputation propagation algorithm that assigns each software agent a multi-dimensional reputation vector rather than a scalar score. Reputation is propagated through an interaction graph via topic-gated transfer operators that modulate each edge by its content embedding, with convergence to a unique fixed point guaranteed by the contraction mapping theorem. We develop a family of Lipschitz-1 transfer operators and composable information-theoretic gates that achieve up to 98% multi-label Precision@5 on dense graphs and 78% on sparse ones. On a benchmark of 50 agents across 8 domains, TrustFlow resists sybil attacks, reputation laundering, and vote rings with at most 4 percentage-point precision impact. Unlike PageRank and Topic-Sensitive PageRank, TrustFlow produces vector reputation that is directly queryable by dot product in the same embedding space as user queries.