Content virality determination and visualization
AssigneeMicrosoft Corporation
InventorsJake Hofman, Robert Gruen, Sharad Goel, Duncan Watts
Life of this patent
Estimating the term…
Abstract
Various techniques of content virality determination and visualization are disclosed herein. For example, a method of determining network content virality metric includes constructing a diffusion cascade for a computer network content based on a plurality of time points at which individual users adopt the network content and connection information of the users. The method also includes calculating, with a processor, a virality metric of the network content based on a structural characteristic of the constructed diffusion cascade. Based on the calculated virality metric, one may determine if the network content is viral.
Claims20
A method of determining network content virality, comprising:
- receiving, with a processor, network data having a plurality of time points at which individual users adopt a network content and connection information of the users in the computer network;
- determining a temporal propagation pattern of the network content across the computer network based on the plurality of time points and the connection information of the users;
- and
- calculating, with the processor, a virality metric of the network content based on a structural characteristic of the propagation pattern.
A method of determining network content virality, comprising:
- determining, with a processor, a propagation pattern for a network content, the propagation pattern representing temporal propagation of the network content from one user to another across a computer network;
- and
- calculating, with the processor, a virality metric of the network content based on a structural characteristic of the constructed propagation pattern.
A computer system for determining network content virality, the computer system having a processor and a computer readable memory containing instructions, when executed by the processor, causing the processor to perform a process comprising:
- constructing a diffusion cascade for a network content based on a plurality of time points at which individual users adopt the network content and connection information of the users, wherein constructing the diffusion cascade include:
- comparing one of the time points to another of the time points;
- determining if the one of the time points corresponds to one of the users connected to another one of the users corresponding to the another of the time points;
- if one of the time points is earlier than another of the time points and the one of the users is connected to the another one of the users,
- designating the one of the users as a parent node and the another one of the users as a child node;
- and
- repeating the comparing, determining, and designating operations until all of the time points are processed to derive the diffusion cascade;
- and
- calculating, with a processor, a virality metric of the network content based on at least one of:
- an average distance between all pairs of the nodes in the diffusion cascade, the distance between a pair of nodes being a distance of shortest path between the pair of nodes;
- a fraction of a largest number of child nodes of a single parent node over a total number of the nodes in the diffusion cascade;
- a probability that two random nodes have a distinct parent node in the diffusion cascade;
- or
- an average depth of the nodes in the diffusion cascade.
Dependencies are inferred from the claim text.