Identifying data patterns
First Claim
1. A computer-implemented method for identifying data patterns from data comprising at least one data object, said data having at least one existing pattern class associated therewith, said data object being represented by a base feature vector, at least one of said base feature vectors having a structure of higher-level and lower-level feature vectors such that at least one element of a higher-level feature vector is a lower-level feature vector;
- said method comprising steps of;
(a) deriving a pattern class prototype in respect of the or each existing pattern class;
(b) creating a new pattern class in respect of a data object; and
(c) deriving a pattern class prototype in respect of said new pattern class,for the or each of said existing pattern classes;
(d1) determining, in respect of elements of a higher-level feature vector, an Element Similarity Measure indicative of a measure of similarity between elements of one or more lower-level feature vectors thereof and corresponding elements of a pattern class prototype of said existing pattern class; and
(d2) determining, in dependence on said Element Similarity Measures, a Vector Similarity Measure indicative of a measure of similarity between said higher-level feature vector and a corresponding vector of a pattern class prototype of said existing pattern class;
(e) determining, in dependence on Vector Similarity Measures determined in respect of each of said existing pattern classes and according to predetermined criteria, whether or not said data object is a member of one of said existing pattern classes;
(f1) in the event of a determination that said data object is a member of an existing pattern class, adding said data object to a set of members of said existing pattern class;
(f2) in the event of a determination that said data object is not a member of an existing pattern class, generating a new pattern class and adding said data object to a set of members of said new pattern class; and
(g) identifying any new pattern class so generated to be indicative of a new data pattern;
wherein one or more of steps (a)-(g) is performed by a computer.
1 Assignment
0 Petitions
Accused Products
Abstract
Methods and systems for identifying data patterns from data comprising at least one data object, said data having at least one existing pattern class associated therewith, said data object being represented by a base feature vector, at least one of said base feature vectors having a structure of higher-level and lower-level feature vectors such that at least one element of a higher-level feature vector is itself a lower-level feature vector.
77 Citations
17 Claims
-
1. A computer-implemented method for identifying data patterns from data comprising at least one data object, said data having at least one existing pattern class associated therewith, said data object being represented by a base feature vector, at least one of said base feature vectors having a structure of higher-level and lower-level feature vectors such that at least one element of a higher-level feature vector is a lower-level feature vector;
- said method comprising steps of;
(a) deriving a pattern class prototype in respect of the or each existing pattern class; (b) creating a new pattern class in respect of a data object; and (c) deriving a pattern class prototype in respect of said new pattern class, for the or each of said existing pattern classes; (d1) determining, in respect of elements of a higher-level feature vector, an Element Similarity Measure indicative of a measure of similarity between elements of one or more lower-level feature vectors thereof and corresponding elements of a pattern class prototype of said existing pattern class; and (d2) determining, in dependence on said Element Similarity Measures, a Vector Similarity Measure indicative of a measure of similarity between said higher-level feature vector and a corresponding vector of a pattern class prototype of said existing pattern class; (e) determining, in dependence on Vector Similarity Measures determined in respect of each of said existing pattern classes and according to predetermined criteria, whether or not said data object is a member of one of said existing pattern classes; (f1) in the event of a determination that said data object is a member of an existing pattern class, adding said data object to a set of members of said existing pattern class; (f2) in the event of a determination that said data object is not a member of an existing pattern class, generating a new pattern class and adding said data object to a set of members of said new pattern class; and (g) identifying any new pattern class so generated to be indicative of a new data pattern; wherein one or more of steps (a)-(g) is performed by a computer. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16)
- said method comprising steps of;
-
17. A system for identifying data patterns from data comprising at least one data object, said data having at least one existing pattern class associated therewith, said data object being represented by a base feature vector, at least one of said base feature vectors having a structure of higher-level and lower-level feature vectors such that at least one element of a higher-level feature vector is a lower-level feature vector;
- said system comprising;
means for deriving a pattern class prototype in respect of the or each existing pattern class; means for creating a new pattern class in respect of a data object; and means for deriving a pattern class prototype in respect of said new pattern class; means for determining, for the or each of said existing pattern classes; an Element Similarity Measure in respect of elements of a higher-level feature vector, said Element Similarity Measure being indicative of a measure of similarity between elements of one or more lower-level feature vectors thereof and corresponding elements of a pattern class prototype of said existing pattern class; and a Vector Similarity Measure in dependence on said Element Similarity Measures, said Vector Similarity Measure being indicative of a measure of similarity between said higher-level feature vector and a corresponding vector of a pattern class prototype of said existing pattern class; means for determining, in dependence on Vector Similarity Measures determined in respect of each of said existing pattern classes and according to predetermined criteria, whether or not said data object is a member of one of said existing pattern classes; means for adding said data object to a set of members of said existing pattern class in the event of a determination that said data object is a member of an existing pattern class; means for generating a new pattern class and adding said data object to a set of members of said new pattern class in the event of a determination that said data object is not a member of an existing pattern class; and means for identifying any new pattern class so generated to be indicative of a new data pattern.
- said system comprising;
Specification