Detecting presence of a subject string in a target string and security event qualification based on prior behavior by an end user of a computer system
View Patent ↗A software-based security agent that hooks into the operating system of a computer device in order to continuously audit the behavior and conduct of the end user of the computer device. The detected actions of the end user can be stored in a queue or log file that can be continuously monitored to detect patterns of behavior that may constitute a policy violation and/or security risk. When a pattern of behavior that may constitute a policy violation and/or security risk is detected, an event may be triggered. A frequency vector string matching algorithm also is disclosed. The frequency vector string matching algorithm may be used to detect the presence or partial presence of subject strings within a target string of alphanumeric characters. The frequency vector string matching algorithm could be used to detect typos in stored computer records or to search for records based on partial information. In addition, the frequency vector string matching algorithm could be used to search communications for sensitive information that has been manipulated, obscured, or partially elided. In addition, an anomaly analysis is disclosed for comparing behavior patterns of one user against the behavior patterns of other users to detect anomalous behaviors.
1. A computer system for detecting presence of a subject string S in a target string T of alphanumeric characters, where T has a length m, the computer system comprising:
a processor circuit; and
a memory in communication with the processor circuit, wherein the memory stores instructions that when executed by the processor circuit cause the processor circuit to determine a similarity score indicative of a similarity between the subject string S and the target string T by:
for each of one or more substrings U of target string T, correlating a frequency of occurrence of a set C of n unique alphanumeric characters between the subject string S and the one or more substrings U, where n≦m;
normalizing a result of the correlation to produce a score for each of the one or more substrings U;
determining the similarity score indicative of the similarity between the subject string S and the target string T by selecting a greatest score from the scores for each of the one or more substrings U; and
at least partially responsive to selecting the greatest score, triggering a response to the subject string S in the target string T.
2. The computer system of claim 1 , wherein:
correlating the frequency of occurrence of the set C of n unique alphanumeric characters between the subject string S and the one or more substrings U for each of one or more substrings U of target string T comprises:
determining a vector ū=[u 1 , u 2 , . . . , u i , . . . , u n ] of n non-negative numbers, wherein the elements of the vector u correspond respectively to a count of the n unique alphanumeric characters in C that are present in a substring U of target string T, wherein the substring U has a length k where n≦k≦m; and
computing a dot product, denoted r, of normalizations of ū and v , where v is a vector of n non-negative numbers, where the elements of the vector v correspond respectively to a count of the unique alphanumeric characters in C that are present in the subject string S; and
normalizing the result of the correlation for each of one or more substrings U of target string T comprises:
computing a dot product, denoted t, of N and v , where N is a vector having n elements all being the same value;
projecting a first interval of values [t, 1] onto a second interval of values [0,1]; and
determining the score for the substring U of string T based on a mapping of r via the projection of the first interval onto the second interval.
3. The computer system of claim 2 , wherein the n elements of vector N if are all 1.
4. A computer-implemented method for detecting presence of a subject string S in a target string T of alphanumeric characters, where T has a length m, the method comprising:
for each of one or more substrings U of target string T:
correlating, by a computer system, a frequency of occurrence of a set C of n unique alphanumeric characters between the subject string S and the one or more substrings U, where n≦m;
normalizing, by the computer system, a result of the correlation to produce a score for each of the one or more substrings U;
determining, by the computer system, a similarity score indicative of the similarity between the subject string S and the target string T by selecting a greatest score from the score for each of the one or more substrings U;
at least partially responsive to selecting the greatest score, triggering a response to the presence of subject string S in the target string T; and
wherein the computer system comprises at least one computer device that comprises a processor circuit and a memory, wherein the memory stores instructions that are executed by the processor circuit.
5. The method of claim 4 , wherein:
correlating the frequency of occurrence of the set C of n unique alphanumeric characters between the subject string S and the one or more substrings U for each of one or more substrings U of target string T comprises:
determining a vector ū=[u 1 , u 2 , . . . , u i , . . . , u n ] of n non-negative numbers, wherein the elements of the vector u correspond respectively to a count of the n unique alphanumeric characters in C that are present in a substring U of target string T, wherein the substring U has a length k where n≦k≦m; and
computing a dot product, denoted r, of normalizations of ū and v , where is a vector of n non-negative numbers, where the elements of the vector v correspond respectively to a count of the unique alphanumeric characters in C that are present in the subject string S; and
normalizing the result of the correlation for each of one or more substrings U of target string T comprises:
computing a dot product, denoted t, of N and v , where N is a vector having n elements all being the same value;
projecting a first interval of values [t, 1] onto a second interval of values [0,1]; and
determining the score for the substring U of string T based on a mapping of r via the projection of the first interval onto the second interval.
6. The method of claim 5 , wherein the n elements of the vector N are all 1.