Abstract: Various embodiments of the technology described herein programmatically expand the capabilities of computing systems to allow for the determination of a host authority value indicative of a reliability of a particular host identified as a node in a Host Navigation Graph (HNG). A classifier or other neural network model is trained, based on the HNG, to classify or determine a reliability the particular host based on the host authority value. Based on the host authority value, the particular host is classified and a mitigation is taken to reduce the negative effects associated with a user interacting with a low-authority host. An example mitigation action includes generating a graphical user interface (GUI) element to notify a user of the reliability of a particular host, for example, before, during, or after a user interacts with the host.

Abstract: Various embodiments of the technology described herein programmatically expand the capabilities of computing systems to allow for the determination of a host authority value indicative of a reliability of a particular host identified as a node in a Host Navigation Graph (HNG). A classifier or other neural network model is trained, based on the HNG, to classify or determine a reliability the particular host based on the host authority value. Based on the host authority value, the particular host is classified and a mitigation is taken to reduce the negative effects associated with a user interacting with a low-authority host. An example mitigation action includes generating a graphical user interface (GUI) element to notify a user of the reliability of a particular host, for example, before, during, or after a user interacts with the host.