Back to results
US 2016/0239763 A1Published application
Motif recognition
AssigneeSecurboration, Inc.
InventorsLee S. Krause, Craig T. Hagan, Bruce R. McQueary, James B. Schneider
Life of this patent
Estimating the term…
Abstract
Methods and systems for motif recognition in graph sets. Distribution scores comparing the canonical representations of a compressed graph in the set of graphs versus the canonical representations of a compressed graph in a reference graph are used to determine the significance of the compressed graph.
Claims19
-20. (canceled)
A method for motif recognition in a set of graphs derived from a collection of noise reduced data, the method comprising:
- compressing an uncompressed graph from the set of graphs resulting in a compressed graph;
- counting the canonical representations of the compressed graph in the set of graphs to yield a first count;
- generating a reference graph from the noise reduced data, the reference graph being equal in size to the uncompressed graph;
- counting the canonical representations of the compressed graph in the reference graph to yield a second count;
- and
- computing a distribution score utilizing at least the first count and the second count to determine the significance of the compressed graph.
3 more elements
A method of classifying at least one social media user, the method comprising:
- detecting at least one motif of the social media user;
- gathering data related to metadata of the social media user;
- gathering data related to social interactions of the social media user;
- computing features of:
- the motif of the user,
- the data related to the metadata of the user, and
- the data related to the social interactions of the user;
- normalizing each computed feature on a zero to one scale;
- aggregating each normalized feature into a numeric vector;
- and
- comparing the numeric vector to at least one model vector, wherein the at least one model vector is representative of a class.
8 more elements
Dependencies are inferred from the claim text.