Extracting information from input data using a semantic cognition network

• US 7,146,380 B2
• Filed: 10/15/2003
• Issued: 12/05/2006
• Est. Priority Date: 10/15/2002
• Status: Active Grant

• Current Assignee: Definiens AG (Astrazeneca PLC)
• Original Assignee: Definiens AG (Astrazeneca PLC)
• Inventors: Binning, Gerd, Schaepe, Arno, Athelogou, Maria, Krug, Christof, Benz, Ursula
• Primary Examiner(s): Cottingham, John R.
• Assistant Examiner(s): LEWIS, CHERYL RENEA
• Application Number:
  US10/687,477
• Publication Number:
  US 20040148296A1
• Time in Patent Office: 1,147 Days
• Field of Search: 706/45, 707103 R-103 Z, 707/100, 707/104.1, 709/218
• US Class Current: 1/1
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition
• CPC Class Codes: G06V 10/248 by interactive preprocessing or interactive shape modelling, e.g. feature points assigned by a user
• CPC Class Codes: G06V 10/422 for representing the structure of the pattern or shape of an object therefor
• CPC Class Codes: G06V 20/69 Microscopic objects, e.g. biological cells or cellular parts
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface
• CPC Class Codes: Y10S 707/99944 Object-oriented database structure
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing
• CPC Class Codes: Y10S 707/99946 Object-oriented database structure network

Abstract

A method is disclosed for extracting information from input data comprising mapping of the input data into a data object network. The method uses a semantic cognition network comprised of the data object network, a class object network and a processing object network. The semantic cognition network uses a set of algorithms to process the semantic units. The semantic cognition network defines a processing object in the processing object network by selecting a data domain in the data object network, a class domain in the class object network and an algorithm from the set of algorithms. The processing object comprises the data domain, the class domain and the algorithm. The processing object is used in the processing object network to process the semantic units.

Claims

1. A computer-implemented method for extracting information from digital image data comprised of semantic units, comprising:

(a) defining a data object network by selecting a plurality of semantic units from said digital image data, wherein a semantic cognition network comprises said data object network, a class object network and a processing object network, wherein said class object network comprises categories of classifications of said selected semantic units, wherein said semantic cognition network is implemented on a distributed network of computers and uses a set of algorithms to process said selected semantic units;

(b) defining a processing object in said processing object network by selecting a data domain in said data object network, a class domain in said class object network and an algorithm from said set of algorithms, wherein said processing object comprises said data domain, said class domain and said algorithm; and

(c) processing said selected semantic units in said processing object network using said processing object, wherein said processing said selected semantic units is used to display a picture region from the digital image data, and wherein the picture region includes semantic units from said data object network that fall within a classification of said class object network.

2. The method of claim 1, wherein said defining in step (b) is performed multiple times, and wherein each of said multiple times occurs in parallel.

3. The method of claim 2, further comprising, between step (b) and step (c):

(d) linking plural processing objects defined when step (b) is performed multiple times, wherein said plural processing objects are linked with each other to form an execution control based on said linking plural processing objects.

4. The method of claim 3, wherein said linking in step (d) comprises at least one of executing said plural processing objects as sub-processing objects, executing said plural processing objects in series or executing said plural processing objects in parallel.

5. The method of claim 1, wherein said processing object is executed several times, each time processing additional semantic units.

6. The method of claim 1, wherein said processing object network comprises a control structure taken from the group consisting of a loop, a branch, a function call, a jump and an execution control.

7. The method of claim 1, wherein said class object network and said processing object network are linked together via said class domain.

8. The method of claim 1, wherein said semantic cognition network uses classifying link objects to process said selected semantic units, and wherein said data object network and said class object network are linked together via said classifying link objects.

9. The method of claim 1, wherein said data object network and said processing object network are linked together via link objects that are dynamically generated during said processing in step (c) based on said data domain.

10. The method of claim 1, wherein one processing object is unambiguously defined by one data domain, one class domain and one algorithm.

11. The method of claim 1, wherein said processing using said processing object in step (c) comprises the steps of:

(c1) extracting a subset of data objects in said data domain from said data object network; and

(c2) applying said algorithm to each of said data object in said subset of data objects.

12. The method of claim 1, wherein said data domain is a subset of said data object network, said class domain is a subset of said class object network, and said algorithm is a subset of said set of algorithms.

13. The method of claim 1, wherein a single processing object processes said selected semantic units in at least one of said data object network, said class object network and said processing object network based on said data domain, said class domain and said algorithm.

14. The method of claim 1, wherein said digital image data is obtained from microscopic pictures taken from tissue samples.

15. The method of claim 1, wherein said digital image data is obtained from a geographical satellite picture.

16. A computer-implemented system for extracting information from input data, comprising:

mapping means for mapping said input data into a data object network, wherein said input data comprises semantic units, wherein a semantic cognition network comprises said data object network, a class object network and a processing object network, wherein said class object network comprises categories of classifications of said semantic units, and wherein said semantic cognition network uses a set of algorithms to process said semantic units;

defining means for defining a processing object in said processing object network by selecting a data domain in said data object network, a class domain in said class object network and an algorithm from said set of algorithms, said processing object comprising said data domain, said class domain and said algorithm;

processing means for processing said semantic units in said processing object network using said processing object; and

display means for displaying a picture region, wherein said picture region is generated using semantic units from said data object network that fall within a classification of said class object network.

17. The system of claim 16, wherein said defining means performs said defining several times, wherein a linking means links said processing object to a second processing object before said processing to form an execution control based on how said processing object is linked to said second processing object.

18. The system of claim 17, wherein said linking means links said processing object to said second processing object by executing said processing objects as sub-processing objects, by executing said processing objects in series or by executing said processing objects in parallel.

19. The system of claim 16, wherein said processing means executes said processing object multiple times.

20. The system of claim 16, wherein said processing object network comprises a control structure taken from the group consisting of a loop, a branch, a function call, a jump and an execution control.

21. The system of claim 16, wherein said class object network and said processing object network are linked together via said class domain.

22. The system of claim 16, wherein said semantic cognition network uses classifying link objects to process said semantic units, and wherein said data object network and said class object network are linked together via said classifying link objects.

23. The system of claim 16, wherein said data object network and said processing object network are linked together via link objects that are dynamically generated based on said data domain.

24. The system of claim 16, wherein one processing object is unambiguously defined by one data domain, one class domain and one algorithm.

25. The system of claim 16, wherein said processing means extracts a subset of data objects in said data domain from said data object network and applies said algorithm to each of said data object in said subset of data objects.

26. The system of claim 16, wherein said data domain is a subset of said data object network, said class domain is a subset of said class object network, and said algorithm is a subset of said set of algorithms.

27. The system of claim 16, wherein a single processing object processes said semantic units in at least one of said data object network, said class object network and said processing object network based on said data domain, said class domain and said algorithm.

28. A system implemented on a computer for extracting knowledge comprising:

a data object network;

a class object network, wherein semantic units are networked to each other in each of said data object network and said class unit network, wherein said semantic units comprise digital image data;

defining means for defining a processing unit by selecting a data domain in said data object network, a class domain in said class object network and an algorithm from a set of algorithms, each of which being part of said processing unit;

processing means for processing said semantic units in accordance with said data domain, said class domain and said algorithm; and

display means for displaying a picture region, wherein said picture region is generated using semantic units processed in accordance with said data domain.

29. A computer-implemented system for extracting information from digital image data, comprising:

a data object network having a data domain, said data object network receiving said digital image data;

a class object network having a class domain;

a processing object network having a processing object, wherein a semantic cognition network comprises said data object network, said class object network and said processing object network, wherein said semantic cognition network uses a set of algorithms to process said digital image data, wherein said processing object comprises said data domain, said class domain and an algorithm from said set of algorithms wherein said processing object network processes said semantic units using said processing object; and

a display on which a region of a digital image is marked to show part of said digital image data that comprises said semantic units processed using said processing object.

30. The system of claim 29, wherein said processing object network comprises a control structure taken from the group consisting of a loop, a branch, a function call, a jump and an execution control.

31. The system of claim 29, wherein said digital image data is obtained from microscopic pictures taken from tissue samples.

32. The system of claim 29, wherein said digital image data is obtained from a geographical satellite picture.