Granted Patent B2

US 10,180,942 · App. 14/087,800 · Granted Jan 15, 2019

Inventors: Igal Raichelgauz (Ramat Gan, IL); Karina Odinaev (Ramat Gan, IL); Yehoshua Y. Zeevi (Haifa, IL)

G06F17/30109G06F17/3002

Abstract

Claims (38)

1. A method for generating concept structures, comprising:

receiving a request to create a new concept structure, wherein the request includes at least a multimedia data element (MMDE) related to the new concept structure;

generating, by a signature generator, at least one signature for the MMDE;

querying a deep-content-classification (DCC) system using the at least one generated signature to find at least one sub-concept, wherein the sub-concept is a concept structure that partially matches the received MMDE, wherein the MMDE is determined to partially match the concept structure when the at least one generated signature and a signature of the concept structure overlap in more than a predetermined threshold level;

checking if the at least one sub-concept satisfies at least one predefined logic rule; and

generating the new concept structure using one or more matching sub-concepts out of the at least one sub-concept that satisfies the at least one predefined logic rule.

2. The method of claim 1 , further comprising:

saving the generated new concept structure in a database communicatively connected to the DCC system.

3. The method of claim 1 , wherein the predefined logic rule indicates a potential relation between the one or more sub-concepts and the new concept structure.

4. The method of claim 1 , wherein the one or more sub-concepts satisfying the at least one predefined logic rule provide sparse representation of the new concept structure.

5. The method of claim 1 , wherein the new concept structure is generated using the one or more sub-concepts if the database does not contain a training set of MMDEs for creation of a new concept structure.

6. The method of claim 1 , wherein the MMDE is at least one of: an image, graphics, a video stream, a video clip, an audio stream, an audio clip, a video frame, a photograph, an image of signals, a medical signal, a geophysical signal, a subsonic signal, a supersonic signal, an electromagnetic signal, and an infrared signal.

7. The method of claim 1 , wherein the DCC system includes:

an attention processor (AP) for generating a plurality of items from the received sensory signal and determining which of the generated items are of interest for signature generation;

the signature generator (SG) for generating at least one signature responsive to at least one item of interest of the sensory signal; and

a concept generator (CG) for matching between the at least one signature generated responsive to at least one item of interest of the sensory signal and a plurality of signature reduced clusters associated with a plurality of cluster structures to identify at least the first set of metadata.

8. A non-transitory computer readable medium having stored thereon instructions for causing one or more processing units to execute the method according to claim 1 .

9. A system for generating concept structures, comprising:

an interface to a network for receiving a request to create a new concept structure, wherein the request includes at least a multimedia data element (MMDE) related to the new concept structure;

a signature generator;

a processor; and

a memory connected to the processor, the memory contains instructions that when executed by the processor, configure the system to:

generate, by the signature generator, at least one signature to the MMDE;

query a deep-content-classification (DCC) system using the at least one generated signature to find at least one sub-concept, wherein the sub-concept is a concept structure that partially matches the received MMDE, wherein the MMDE is determined to partially match the concept structure when the at least one generated signature and a signature of the concept structure overlap in more than a predetermined threshold level;

check if the at least one sub-concept satisfies at least one predefined logic rule; and

generate the new concept structure using one or more matching sub-concepts out of the at least one sub-concept that satisfies the at least one predefined logic rule.

10. The system of claim 9 , wherein the system is further configured to:

save the generated new concept structure in a database communicatively connected to the DCC system.

11. The system of claim 10 , wherein the predefined logic rule indicates a potential relation between the one or more sub-concepts and the new concept structure.

12. The system of claim 9 , wherein the one or more sub-concepts satisfying the at least one predefined logic rule provide sparse representation of the new concept structure.

13. The system of claim 9 , wherein the new concept structure is generated using the sub-concepts if the database does not contain a training set of MMDEs for creation of the concept structure.

14. The system of claim 9 , wherein the MMDE is at least one of: an image, graphics, a video stream, a video clip, an audio stream, an audio clip, a video frame, a photograph, an image of signals, a medical signal, a geophysical signal, a subsonic signal, a supersonic signal, an electromagnetic signal, and an infrared signal.

15. The system of claim 9 , wherein the DCC system includes:

an attention processor (AP) for generating a plurality of items from the received sensory signal and determining which of the generated items are of interest for signature generation;

the signature generator (SG) for generating at least one signature responsive to at least one item of interest of the sensory signal; and

a concept generator (CG) for matching between the at least one signature generated responsive to at least one item of interest of the sensory signal and a plurality of signature reduced clusters associated with a plurality of cluster structures to identify at least the first set of metadata.

16. The method of claim 1 , wherein the MMDE is determined to partially match the concept structure when the at least one generated signature and a signature of the concept structure overlap less than a second predetermined threshold level which when at least equaled indicates that the MMDE fully matches the concept structure.

17. The system of claim 9 , wherein the MMDE is determined to partially match the concept structure when the at least one generated signature and a signature of the concept structure overlap less than a second predetermined threshold level which when at least equaled indicates that the MMDE fully matches the concept structure.

Assignments (3)

Priority Claims (2)

Continuity (13)