Assessing the Stability of Rankings in Knowledge Graphs Against Perturbations

KG ranking vulnerability to perturbations: Crowdsourced knowledge graphs are vulnerable to intentional or unintentional perturbations that can significantly affect rankings derived from the graphs.

Structural ranking stability assessment: The paper addresses the need to understand how relationship-level perturbations affect the stability of entity rankings in knowledge graphs.

Entity-level perturbation focus gap: Prior work has mainly focused on detecting and preventing entity-level perturbations rather than assessing structural ranking stability.

KG ranking vulnerability to perturbations: Crowdsourced knowledge graphs are vulnerable to intentional or unintentional perturbations that can significantly affect rankings derived from the graphs.

Structural ranking stability assessment: The paper addresses the need to understand how relationship-level perturbations affect the stability of entity rankings in knowledge graphs.

Ranking stability formalization: The paper formalizes the problem of ranking stability under perturbations in knowledge graphs.

Probabilistic rank-change model: The paper proposes a probabilistic model to assess whether modifications to knowledge graph relationships are likely to change entity ranks.

Complex network vulnerability analysis: The paper uses complex network analysis to evaluate vulnerabilities of rankings under perturbations.

Ranking stability formalization: The paper formalizes the problem of ranking stability under perturbations in knowledge graphs.

Probabilistic rank-change model: The paper proposes a probabilistic model to assess whether modifications to knowledge graph relationships are likely to change entity ranks.

Perturbation-dependent ranking resilience:

Perturbation-dependent ranking resilience: