Dynamic multidimensional risk-weighted suspicious activities detector
A computerized method is established to detect suspicious and fraudulent activities in a group of subjects by defining and dynamically integrating multidimensional risks, which are based on the characteristics of the subjects, into a mathematical model to produce a set of the most up-to-date representative risk values for each subject based on its activities and background. These multidimensional risk definitions and representative risk values are used to select a subset of multidimensional risk-weighted detection algorithms so that suspicious or fraudulent activities in the group of subjects can be effectively detected with higher resolution and accuracy. A priority sequence, which is based on the set of detection algorithms that detect the subject and the representative risk values of the detected subject, is produced to determine the priority of each detected case during the investigation process. To assist the user to make a more objective decision, any set of multidimensional risks can be used to identify a group of subjects that contain this set of multidimensional risks so that group statistics can be obtained for comparison and other analytical purposes. Furthermore, to fine-tune the system for future detections and analyses, the detection results are used as the feedback to adjust the definitions of the multidimensional risks and their values, the mathematical model, and the multidimensional risk-weighted detection algorithms.
1 . A computer assisted method to facilitate the detection of suspicious and fraudulent activities in a group of subjects by:
storing activities and background data associated with each subject into a database;
defining multidimensional risks based on said activities and background data;
assigning a risk value to each of the multidimensional risks;
analyzing the database to determine a respective set of multidimensional risk values for each subject;
establishing a mathematical model which reduces each set of multidimensional risk values to one or more representative risk values which represent the overall risks of the respective subject;
applying the mathematical model to each said set of multidimensional risk values to thereby produce a set of representative risk values for each subject,
establishing a set of multidimensional risk-weighted detection algorithms that include said representative risk values;
using the representative risk values for each subject to select a respective subset of said multidimensional risk-weighted detection algorithms applicable to that particular subject; and
identifying potentially suspicious or fraudulent activities by using said respective subset of said multidimensional risk-weighted detection algorithms to analyze and detect suspicious or fraudulent activities associated with said particular subject.
2 . The method of claim 1 wherein:
the analyzing database to determine a respective set of multidimensional risk values for each subject and applying mathematical model to each said set of multidimensional risk values to thereby produce a set of representative risk values for each subject steps are repeated to thereby dynamically update said respective risk values prior to performing the identifying step on more recent activities.
3 . The method of claim 1 wherein: the multidimensional risks include at least the possible transactional patterns of the subjects.
4 . The method of claim 1 wherein:
the multidimensional risks include at least the possible behavior patterns of the subjects.
5 . The method of claim 1 wherein:
the multidimensional risks include at least the possible historical patterns of the subjects.
6 . The method of claim 1 wherein:
the multidimensional risks include at least the possible natures of the subjects.
7 . The method of claim 1 wherein:
the multidimensional risks include at least the possible geographical locations of the subjects.
8 . The method of claim 1 wherein:
the multidimensional risks include at least the possible social status of the subjects.
9 . The method of claim 1 wherein:
the multidimensional risks include at least the possible business types of the subjects.
10 . The method of claim 1 wherein:
the multidimensional risks include at least the possible occupation types of the subjects.
11 . The method of claim 1 wherein:
the multidimensional risks include at least the possible identification codes of the subjects.
12 . The method of claim 1 wherein:
the multidimensional risks include at least the possible political relationships of the subjects.
13 . The method of claim 1 wherein:
the multidimensional risks include at least the possible foreign activities of the subjects.
14 . The method of claim 1 wherein:
the multidimensional risks include at least the possible ownership categories of the subjects.
15 . The method of claim 1 wherein:
the multidimensional risks include at least the possible organizational structures of the subjects.
16 . The method of claim 1 wherein:
the mathematical model at least involves addition of two or more risk values.
17 . The method of claim 1 wherein:
the mathematical model at least involves subtraction of a first risk value from a second risk value.
18 . The method of claim 1 wherein:
the mathematical model at least involves multiplication of a first risk value by a second risk value.
19 . The method of claim 1 wherein:
the mathematical model at least involves division of a first risk value by a second risk value.
20 . The method of claim 1 wherein:
the mathematical model at least involves a polynomial function.
21 . The method of claim 1 wherein:
the mathematical model at least involves an exponential function.
22 . The method of claim 1 wherein:
the mathematical model at least involves a logarithm function.
23 . The method of claim 1 wherein:
the mathematical model at least involves a trigonometric function.
24 . The method of claim 1 wherein:
the mathematical model at least involves an inverse trigonometric function.
25 . The method of claim 1 wherein:
the mathematical model at least involves a linear transformation.
26 . The method of claim 1 wherein:
the mathematical model at least involves a non-linear transformation.
27 . The method of claim 1 wherein:
the set of representative values includes at least a single value.
28 . The method of claim 1 further comprising:
the set of representative risk values of a subject is produced from the mathematical model based on the set of multidimensional risk values of the subject dynamically so that this set of representative risk values precisely represents the most up-to-date characteristics of the possibly evolving activities or background of the subject.
29 . The method of claim 1 further comprising:
adjusting the definitions of the multidimensional risks and their values manually based on the feedback from the detection results of suspicious or fraudulent activities.
30 . The method of claim 1 further comprising:
adjusting the definitions of the multidimensional risks and their values automatically based on the feedback from the detection results of suspicious or fraudulent activities.
31 . The method of claim 1 further comprising:
adjusting the mathematical model manually based on the feedback from the detection results of suspicious or fraudulent activities.
32 . The method of claim 1 further comprising:
adjusting the mathematical model automatically based on the feedback from the detection results of suspicious or fraudulent activities.
33 . The method of claim 1 further comprising:
adjusting the multidimensional risk-weighted detection algorithms manually based on the feedback from the detection results of suspicious or fraudulent activities.
34 . The method of claim 1 further comprising:
adjusting the multidimensional risk-weighted detection algorithms automatically based on the feedback from the detection results of suspicious or fraudulent activities.
35 . A computerized method to define a comprehensive set of multidimensional risks and their values, and to assign a set of risk values to each of the subjects in a database, by
identifying all the possible different risk dimensions associated with the group of subjects based on the activities and background of these subjects;
grouping a similar type of risk dimensions into the same risk category;
using the same definition to describe the common part of all risk dimensions in the same risk category but including at least one variable so that a “multidimensional risk template” is produced for this particular risk category;
producing a separate “multidimensional risk template” for each risk category;
permitting the user to enter a different value for each variable into multiple copies of the same “multidimensional risk template” to complete the definition of multiple risk dimensions within the same risk category;
saving the defined risk dimensions and their associated risk values into the database;
using the definition of a risk dimension and its risk value to produce a database program, which will assign the said risk value to any subject in the database that has a risk exposure matching the definition of the risk dimension; and
repeating the preceding step for each of the risk dimensions until the entire set of multidimensional risks and their values have been assigned to all the subjects in the database.
36 . The method of claim 35 further comprising:
choosing any of the risk dimensions and identifying all the subjects that contain such a risk dimension so that analyses can be performed for this specific group of subjects that contain such a risk.
37 . The method of claim 36 further comprising:
choosing any individual subject and identifying a specific group of subjects that contain a specific risk dimension as the individual subject; and
comparing the patterns of such an individual subject with the statistical patterns of such a group to identify possible suspicious and fraudulent activities of such an individual subject.
38 . The method of claim 35 further comprising:
choosing any set of the risk dimensions and identifying all the subjects that contain such a set of risk dimensions so that analyses can be performed for this specific group of subjects that contain such a set of risks.
39 . The method of claim 38 further comprising:
choosing any individual subject and identifying a specific group of subjects that contain a specific set of risk dimensions as the individual subject; and
comparing the patterns of such an individual subject with the statistical patterns of such a group to identify possible suspicious and fraudulent activities of such an individual subject.
40 . A computer assisted method to facilitate the detection of suspicious and fraudulent activities in a group of subjects by:
storing activities and background data associated with each subject into a database;
defining multidimensional risks based on said activities and background data;
assigning a risk value to each of the multidimensional risks;
analyzing the database to determine a respective set of multidimensional risk values for each subject;
establishing a mathematical model which reduces each set of multidimensional risk values to one or more representative risk values which represent the overall risks of the respective subject;
applying the mathematical model to each said set of multidimensional risk values to thereby produce a set of representative risk values for each subject,
establishing a set of multidimensional risk-weighted detection algorithms that can be selected based on the multidimensional risks;
using the multidimensional risks of each subject to select a respective subset of said multidimensional risk-weighted detection algorithms applicable to that particular subject; and
identifying potentially suspicious or fraudulent activities by using said respective subset of said multidimensional risk-weighted detection algorithms to analyze and detect suspicious or fraudulent activities associated with said particular subject.
41 . A computerized method to dynamically produce a set of representative risk values for each of the subjects in a database by:
defining a comprehensive set of multidimensional risks and their values based on the activities and background of the subjects;
assigning the set of risk values to each of the subjects based on the characteristics of the subject;
establishing a mathematical model based on the set of multidimensional risks;
using the mathematical model to produce a set of representative risk values for each subject in the database; and
repeating the assigning risk values and using mathematical model steps dynamically so that the set of representative risk values of each subject represents the updated overall risks of each of the subjects in the database.
42 . A computerized method to establish the priority of investigating cases of suspicious activities, which are detected by Multidimensional Risk-Weighted Detections, for a group of subjects in a database by
identifying a set of representative risk values that represent the overall risks for each of the subjects based on the multidimensional risks;
identifying all the possible detection algorithms based on the purpose of the system;
assigning a priority value to each of the detection algorithms,
combining the priority values of those detection algorithms that jointly detect a subject with the representative risk values of said detected subject to form a decision vector; and
determining whether said detected case should be investigated with higher priority based on the characteristics of this decision vector.
43 . The method of claim 42 wherein:
the characteristics of the decision vector are measured at least through a mathematical transformation of the decision vector.
44 . The method of claim 43 wherein:
the mathematical transformation includes at least the proportional adjustment of the value of each component of the vector to make the maximum value of each component of the vector about the same magnitude (a “normalization” process to “normalize” a vector).
45 . The method of claim 43 wherein:
the mathematical transformation includes at least the root square of the summation of the square of each component of the normalized vector.
46 . The method of claim 43 wherein:
the mathematical transformation includes at least the summation of the square of each component of the normalized vector.
47 . The method of claim 43 wherein:
the mathematical transformation includes at least the summation of the components of the normalized vector.
48 . The method of claim 43 wherein:
the mathematical transformation includes at least the summation of the components of the decision vector.
49 . The method of claim 43 wherein:
the mathematical transformation includes at least the combining of the components of the decision vector into a single value through a mathematical model.
50 . The method of claim 43 wherein:
the mathematical transformation includes at least the combining of the components of the normalized decision vector into a single value through a mathematical model.