Systems and methods for determining features of entities based on centrality metrics of the entities in a knowledge graph

A method of determining network-based feature propagation in a network of entities stored in a graph database, the method implemented on a computer system having a processor programmed to perform the method, the method comprising:

• identifying, by a computer system, an entity that is associated with a feature of interest, the feature of interest being assigned with a quantitative feature value;
• extracting, by the computer system, a subgraph from the graph database based on the identified entity, the subgraph comprising a plurality of entities including the identified entity and one or more neighbors of the identified entity;
• generating, by the computer system, an adjacency list that describes relationships between the plurality of entities in the subgraph;
• generating, by the computer system, a plurality of centrality metrics based on the adjacency list, each centrality metric indicating a level of influence an entity in the subgraph has on other entities in the subgraph based on a level of connectedness between the entity and the other entities;
• determining, by the computer system, a network effect of the identified entity on the one or more neighbors based on the plurality of centrality metrics and types of relationships between the plurality of entities in the subgraph;
• and
• generating, by the computer system, a respective quantitative feature value for each neighbor based on the determined network effects and the quantitative feature value.

A method, comprising:

• receiving, by a processor, a request to assess an entity, the request comprising an identification of the entity;
• identifying, by the processor, a first entity of interest that is associated with a feature of interest;
• identifying, by the processor, one or more paths from the entity to the first entity of interest, the one or more paths, each path comprising one or more entities in the path;
• determining, by the processor, a plurality of path lengths, each path length corresponding to a respective path of the one or more paths;
• determining, by the processor, a plurality of centrality metrics for each entity in each path, each centrality metric indicating a level of influence an entity in a subgraph has on other entities in the subgraph based on a level of connectedness between the entity and the other entities;
• and
• determining, by the processor, a network effect of the first entity of interest on the entity based on the plurality of centrality metrics and the plurality of path lengths.