Identifying related objects using quantum clustering
AssigneeOrcatec LLC
InventorsHerbert L. Roitblat, Brian Golbére
Life of this patent
Estimating the term…
Abstract
Techniques for grouping related objects such as documents and files using quantum clustering are disclosed. A method may include constructing a feature-object database of multiple objects. The feature-object database may have quantized selected features as keys. A connected objects database maybe built. Clusters of connected objects may be identified in the connected objects database. The clusters of identified objects may be evaluated to determine groups of related objects. The method may be implemented on a computing device.
Claims17
A computing device having a processor and accessible computer readable storage media, the storage media having instructions thereon that when executed by the processor cause the computing device to perform a method for relating objects, the method comprising:
- forming a directed graph of connected objects from a feature-object database,
- wherein the feature-object database comprises,
- a plurality of feature-object database keys, each feature-object database key comprising a feature value calculated from a feature extracted from an object;
- and
- at least one feature-object database value corresponding to each feature-object database key, the at least one feature-object database value comprising an object identifier identifying an object having the feature corresponding to the corresponding feature-object database key;
- and
- wherein the forming a directed graph comprises,
- selecting keys from the feature object database, the selected keys having a plurality corresponding feature-object database values, and for each selected key, selecting the corresponding feature-object database values and adding the corresponding object identifiers as connected nodes in the directed graph of connected objects;
- and
- calculating the number of shared features between pairs of objects, wherein a magnitude of an edge between two connected nodes is related to the number of shared features between the two objects represented by the connected nodes;
- and
- identifying clusters each comprising a plurality of connected objects from the directed graph of connected objects.
A storage medium having instructions stored thereon which when executed by a processor cause the processor to perform the actions comprising:
- forming a directed graph of connected objects from a feature-object database,
- wherein the feature-object database comprises,
- a plurality of feature-object database keys, each feature-object database key comprising a feature value calculated from a feature extracted from an object;
- and
- at least one feature-object database value corresponding to each feature-object database key, the at least one feature-object database value comprising an object identifier identifying an object having the feature corresponding to the corresponding feature-object database key;
- and
- wherein the forming a directed graph comprises,
- selecting keys from the feature object database, the selected keys having a plurality corresponding feature-object database values, and for each selected key selecting the corresponding feature-object database values and adding the corresponding object identifiers as connected nodes in the directed graph of connected objects;
- and
- calculating the number of shared features between pairs of objects,
- wherein a magnitude of an edge between two connected nodes is related to the number of shared features between the two objects represented by the connected nodes;
- and
- identifying clusters each comprising a plurality of connected objects from the directed graph of connected objects.
The storage medium of 12 wherein the calculating the number of share features comprises incrementing a pair-count for each pair of object identifiers appearing as selected feature-object database values.
Dependencies are inferred from the claim text.