Method and system for finding conjunctive clusters

1. A computer implemented method for finding conjunctive clusters comprising:

identifying a first group of points, wherein said first group of points is electronic data;

identifying a second group of points, wherein said second group of points is electronic data;

randomly drawing a first sample from said first group of points of a first predetermined size; and

identifying subsets of said first sample;

for each subset of said first sample of a second predetermined size perform steps comprising;

identifying a subgroup of said second group of points that share an intersection with all of the points in said first sample of points;

identifying a subgroup of said first group of points that share an intersection with a specified number of the points in said subgroup of said second group of points; and

outputting a third group of points wherein the product of the magnitude of said subgroup of said second group and the magnitude of said subgroup of said first group is maximized, wherein one or more conjunctive clusters are found.

2. The computer implemented method of claim 1 , further comprising drawing a second sample from said first group of points wherein said subgroup of said first group of points is identified from said second sample.

3. The computer implemented method of claim 2 , wherein said second group of points represent attributes.

4. The computer implemented method of claim 3 , wherein said third group represents a biclique or conjunctive cluster.

5. The computer implemented method of claim 4 , wherein said conjunctive cluster represents overlapping clusters.

6. The computer implemented method of claim 5 , wherein each of said points belongs to multiple clusters.

7. The computer implemented method of claim 6 , wherein each of said attributes is used to describe multiple clusters.

8. The computer implemented method of claim 7 , wherein a point is not clustered if it does not fall into a specified number of largest clusters in a set of clusters.

9. The computer implemented method of claim 8 , wherein an entire dataset is not read into a main memory.

10. The computer implemented method of claim 4 further comprising the step of:

collecting a plurality of bicliques.

11. A computer useable storage medium having computer useable code embodied therein causing a computer to perform operations comprising:

determining a first group of points in a conjunctive clustering process;

determining a second group of points in a conjunctive clustering;

randomly selecting a first sample from said first group of points of a first predetermined size; and

identifying subsets of said first sample;

for each subset of said first sample of a second predetermined size perform steps comprising:

determining a subgroup of said second group of points that share an intersection with all of the points in said first sample of points;

determining a subgroup of said first group of points that share an intersection with a specified number of the points in said subgroup of said second group of points; and

outputting a third group of points wherein the product of the magnitude of said subgroup of said second group and the magnitude of said subgroup of said first group is maximized, wherein one or more conjunctive clusters are found.

12. The computer useable storage medium of claim 11 , further comprising drawing a second sample from said first group of points wherein said subgroup of said first group of points is identified from said second sample.

13. The computer useable storage medium of claim 12 , wherein said second group of points represent attributes.

14. The computer useable storage medium of claim 13 , wherein said third group represents a biclique or conjunctive cluster.

15. The computer useable storage medium of claim 14 , wherein said conjunctive cluster represents overlapping clusters.

16. The computer useable storage medium of claim 15 , wherein each of said points belongs to multiple clusters.

17. The computer useable storage medium of claim 16 , wherein each of said attributes is used to describe multiple clusters.

18. The computer useable storage medium of claim 17 , wherein a point is not clustered if it does not fall into a specified number of largest clusters in a set of clusters.

19. The computer useable storage medium of claim 18 , wherein an entire dataset is not read into a main memory.

20. The computer useable storage medium of claim 14 further comprising the step of:

collecting a plurality of bicliques.

21. A computer system comprising:

a processor;

a memory device comprising computer executable instructions stored therein for performing a method for finding conjunctive cluster comprising:

ascertaining a first group of points;

ascertaining a second group of points;

randomly choosing a first sample from said first group of points of a first predetermined size; and

identifying subsets of said first sample;

for each subset of a second predetermined size of said first sample perform steps comprising;

ascertaining a subgroup of said second group of points that share an intersection with all of the points in said subset;

ascertaining a subgroup of said first group of points that share an intersection with a specified number of the points in said subgroup of said second group of points; and

outputting a third group of points wherein the product of the magnitude of said subgroup of said second group and the magnitude of said subgroup of said first group is maximized, wherein one or more conjunctive clusters are found.

22. The computer system of claim 21 , further comprising drawing a second sample from said first group of points wherein said subgroup of said first group of points is identified from said second sample.

23. The computer system of claim 22 , wherein said first group of points represent attributes.

24. The computer system of claim 23 , wherein said third group represents a biclique or conjunctive cluster.

25. The computer system of claim 24 , wherein said conjunctive cluster represents overlapping clusters.

26. The computer system of claim 25 , wherein each of said points belongs to multiple clusters.

27. The computer system of claim 26 , wherein each of said attributes is used to describe multiple clusters.

28. The computer system of claim 27 , wherein a point is not clustered if it does not fall into a specified number of largest clusters in a set of clusters and wherein an entire dataset is not read into main memory.

29. The computer system of claim 24 further comprising:

collecting a plurality of bicliques.

30. A computer implemented method for finding conjunctive clusters comprising:

(a) monitoring a stream sequence of points, wherein the stream sequence of points is electronic data;

(b) identifying a series of discrete portions of said sequence of points and for each discrete portion of said sequence of points:

(c) identifying a first group of points;

(d) randomly drawing a first sample of points from said first group of points of a specified size;

(e) identifying subsets of said first sample of points of a specified size; and

(f) initializing a first plurality of counts of the number of points from said streaming sequence of points that satisfy the attributes in each of said plurality of counts of the number of points from said streaming sequence of points that satisfy the attributes in each of said subsets of said first sample, and an additional attribute from said first group,

wherein for each point in a first discrete portion of said sequence of points that satisfies the attributes in each subset in said first sample perform steps comprising:

(g) incrementing said first plurality of counts based on said streaming sequence of points;

(h) incrementing said second plurality of counts based on said streaming sequence of points;

(i) outputting a third group of points that constitute the largest biclique based on the first and second plurality of counts;

(j) for the next discrete portion of said sequence of points performing steps (g) through (i), wherein one or more conjunctive clusters are found.