Access management using thread of bytes
By generating cluster threads from historical access requests and matching real-time requests, the method addresses the inefficiencies of high-dimensional input detection, ensuring accurate and efficient access request validation.
Patent Information
- Application Number
- PCT/US2025/028000
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2025-05-06
- Publication Date
- 2026-11-12
AI Technical Summary
Existing methods for detecting potentially invalid access requests in high-dimensional input scenarios are computationally expensive and resource-intensive, especially when handling millions of requests per minute, necessitating a more efficient detection mechanism.
A method that generates cluster threads based on historical access requests, determines access metrics, and matches real-time requests to these threads to determine access outcomes, reducing computational overhead while maintaining accuracy.
This approach efficiently handles high-dimensional inputs by organizing and condensing complex data, reducing scalability issues and maintaining speed and performance without compromising accuracy.
Smart Images

Figure US2025028000_12112026_PF_FP_ABST
Abstract
Description
PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01ACCESS MANAGEMENT USING THREAD OF BYTES BACKGROUND
[0001] Detecting potentially invalid access requests requires analyzing certain individual data elements or features associated with an access request. Often, having more individual data elements or features available for analysis increases the accuracy of detecting potentially invalid access requests in some situations. However, handling high-dimensional input is computationally expensive and resource intensive, especially at scale when millions of access requests occur every minute. Accordingly, there is a need for a detection mechanism that can handle high-dimensional inputs efficiently.
[0002] Embodiments of the disclosure address this problem and other problems individually and collectively.BRIEF SUMMARY
[0003] Embodiments can secure access to a resource using a set of historical access requests from a plurality of devices. Each access request includes a resource identifier and detector values. A cluster thread is generated for each access request based on its detector values. For each cluster thread, one or more access metrics is determined using received access result values received. Embodiments can further includes receiving real-time access requests, generating a current cluster thread for the real-time access request, and matching the current cluster thread to a cluster thread in the set of cluster threads generated for the historical access requests. Based on the access metrics of the matching cluster thread, an access request outcome is determined and access to the resource is provided based on the determined access request outcome.
[0004] Certain embodiments of the present disclosure can provide methods, systems, and apparatuses for identifying potentially invalid access requests. In an example, a method for securing access to resources includes receiving, over a network from a plurality of devices, a set of access requests, each including a resource identifier for a respective resource and a plurality of detector values corresponding to at least three fields, the set of access requests including at least 500,000 access requests. The method also includes storing access result values for the set of access requests, where the access result values provide a measure of 179088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 approval in granting valid access to the respective resource. The method also includes selecting a first field of the at least three fields to be a first seed field. The method also includes for each particular detector value of the first seed field: determining a category' label for other fields of the at least three fields, where the category label indicates an amount of different detector values in access requests that include the particular detector value of the first seed field; generating a segment thread of category labels for the particular detector value, thereby generating a cluster thread. The method also includes for each cluster thread of a set of cluster threads, determining one or more access metrics from the access result values. The method also includes receiving, over a network from a plurality' of devices, a plurality of real-time access requests. The method also includes for each real-time access request: extracting (1) a current detector value for the first seed field and (2) the resource identifier for the respective resource; generating, using the current detector value, a current cluster thread of category labels from the set of access requests; matching the current cluster thread to a first cluster thread of the set of cluster threads; determining an access request outcome for the real-time access request using the one or more access metrics of the first cluster thread; and providing access to the resource based on the access request outcome.
[0005] Another example includes a method for securing access to resources. The method also includes receiving, over a network from a plurality of devices, a set of access requests, each including a resource identifier for a respective resource and a plurality' of geographical detector values corresponding to at least three fields, the set of access requests including at least 500.000 access requests. The method also includes storing access result values for the set of access requests, where the access result values provide a measure of approval in granting valid access to the respective resource. The method also includes for each access request of the set of access requests: performing a pairwise comparison between each pair of the plurality’ of geographical detectors, where each pairwise comparison provides a category label indicating whether a match exists between the pair of geographical detectors, thereby generating a set of segment threads of category labels; and generating a cluster thread from the set of segment threads of category’ labels. The method also includes for each cluster thread of a set of cluster threads, determining one or more access metrics from the access result values. The method also includes receiving, over a network from a plurality of devices, a plurality' of real-time access requests. The method also includes for each real-time access request: extracting (1) the plurality' of geographical detector values and (2) the resource 279088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 identifier for the respective resource; generating, using the plurality of geographical detector values, a current cluster thread of category labels; matching the current thread to a first cluster thread of the set of cluster threads; determining an access request outcome for the real-time access request using the one or more access metrics of the first cluster thread; and providing access to the resource based on the access request outcome.
[0006] Some embodiments of the present disclosure include a computer system for securing access to resources, the computer system may include: one or more processors; a non-transitory computer readable medium storing a plurality' of instructions that, when executed, control the one or more processors to perform the method of any of the methods described herein. Implementations of the described techniques may include hardware, a method or process, or computer software on a computer-accessible medium.
[0007] These and other embodiments of the disclosure are described in detail below. For example, other embodiments are directed to systems, devices, and computer readable media associated with methods described herein.
[0008] A better understanding of the nature and advantages of embodiments of the present disclosure may be gained with reference to the following detailed description and the accompanying drawings.BRIEF DESCRIPTION OF THE DRAWINGS
[0009] FIG. 1 shows a resource security system for authorizing access to resources, in accordance with some embodiments of the present invention.
[0010] FIG. 2A shows a graph depicting the relationship between accuracy of detecting potentially invalid access requests and number of dimensions, in accordance with some embodiments of the present invention.
[0011] FIG. 2B shows a graph depicting the relationship between the risk index increment and number of dimensions, in accordance with some embodiments of the present invention.
[0012] FIG. 3A shows a block diagram of an exemplary detection mechanism for generating an index of historical access request outcomes, in accordance with some embodiments of the present invention.379088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01
[0013] FIG. 3B shows a block diagram of an exemplary detection mechanism for generating an outcome for a current access request, in accordance with some embodiments of the present invention.
[0014] FIG. 4A shows an exemplary segment thread generated from cross combination morphing detectors, in accordance with some embodiments of the present invention.
[0015] FIG. 4B shows an exemplary' historical access result index generated from cross combination morphing detectors, in accordance with some embodiments of the present invention.
[0016] FIG. 5A shows a flowchart of a method for securing access to resources by determining the validity of an access request, in accordance with some embodiments of the present invention.
[0017] FIG. 5B is a continuation of the flowchart depicted in FIG. 5A.
[0018] FIG. 6A shows exemplary' geographical detector vectors generated from combination geographical detectors, in accordance with some embodiments of the present invention.
[0019] FIG. 6B shows an exemplary geographical detector thread, in accordance with some embodiments of the present invention.
[0020] FIG. 7 shows an exemplary' historical access result index generated from combination geographical detectors, in accordance with some embodiments of the present invention.
[0021] FIG. 8A shows a flowchart of a method for securing access to resources by determining the validity of an access request, in accordance with some embodiments of the present invention.
[0022] FIG. 8B is a continuation of the flowchart depicted in FIG. 8A.
[0023] FIG. 9 shows a block diagram of an exemplary computer apparatus, in accordance with some embodiments of the present invention.479088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 TERMS
[0024] Prior to discussing embodiments of the disclosure, description of some terms may be helpful in understanding embodiments of the disclosure.
[0025] The term ‘‘resource” can refer to any asset that may be used or consumed. For example, the resource may be an electronic resource (e.g.. stored data, received data, a computer account, a network-based account, an email inbox), a physical resource (e g., a tangible object, a building, a safe, or a physical location), or other electronic communications between computers (e.g., a communication signal corresponding to an account for performing a transaction). Thus, a physical resource can be a physical object.
[0026] The term “access request” (also referred to as an “authentication request”) can refer to a request to access a resource. The access request may be received from a requesting computer, a user device, or a resource computer, for example. The access request may include authentication information (also referred to as authorization information), such as a user name, resource identifier, or password. The access request may also include and access request parameters, such as an access request identifier, a resource identifier, a timestamp, a date, a device or computer identifier, a geo-location, or any other suitable information. A real-time access request occurs when access to a resource is desired at the time the request is made. In such situations, it is desirable to provide a quick determination for whether to provide access to the resource.
[0027] The term “access result values” can refer to possible outcomes of an access request, e.g., acceptances, rejections, fraud, and review. The set of outcomes determine whether or not access to the resource was granted to a user. The access result values can be counted to provide a set of counts of access results. As examples, the access result values can comprise a count for accepted access results, a count for rejected access results, a count for invalid access results, and a count for access results that need to be reviewed. The access result values may be included wi thin historical access result data and used by a detector to generate a measure of approval for an access request.
[0028] The term “field” can refer to a label given to a data element or a combination of data elements extracted from an access request. The label refers to the category the data579088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 element or combination of data elements belongs to (e.g., “email’', “user_id”, “account number”, etc.).
[0029] The term “detector” can refer to any suitable structure or element that can evaluate a data element or a combination of data elements extracted from an access request. An evaluated data element may comprise a user name, an email, an account number, a password, etc. For a detector that evaluates a combination of data elements, exemplary' combinations of data elements may include the user name and the email; the email and the account number; or the user name, the email, and the account number. A set of detectors can be used to generate a score measuring the legitimacy of an access request based upon historical access result data.
[0030] The term “seed” or “seed detector” can refer to a detector chosen from a set of detectors to act as the reference value for the other detectors. Based on the seed or seed detector, a category label can be assigned to each detector in the set of detectors based on values of the detectors (seed detector and other detectors) relative to the value of the seed vector. For example, the category label can be determined based on the number of unique detector values determined for that detector associated with the seed or seed detector. As another example, a category label can correspond to whether another detector has a same value as the seed detector. Each seed, or seed detector, can be associated with a segment thread.
[0031] The term “segment thread”, “segment”, or “section” can refer to a collection of category labels assigned to the detectors in a set of detectors.
[0032] The term “thread” can refer to a collection of segment threads. However, in the instance where only one seed detector is used, then the segment thread is equivalent to the thread and vice versa.
[0033] The term “cluster thread” can refer to a collection of thread(s) generated for each particular seed detector in the collection of seed detector(s) identified for a particular access request.
[0034] The term “dimension” can refer to a number of parameters, fields, or variables used to determine whether an access request is valid.
[0035] The term “historical access result data” can refer to a collection of access results associated with a detector. Historical access result data may include all of the elements of 679088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 each of the access results, such as authentication information (e.g., a user name, resource identifier, or password) and access request parameters (e.g., access request identifier, a resource identifier, a timestamp, a date, a device or computer identifier, a geo-location, or any other suitable information). Furthermore, the historical access result data may comprise information (e.g., an evaluation score) about whether or not a past access result was accepted or rejected. In some embodiments, the historical access result data may only comprise a set of access results for a predetermined time period. A set of access results may also further be compartmentalized into shorter time windows within the predetermined time period.
[0036] The term “server computer” may include a powerful computer or cluster of computers. For example, the server computer can be a large mainframe, a minicomputer cluster, or a group of computers functioning as a unit. In one example, the server computer may be a database server coupled to a web server. The server computer may be coupled to a database and may include any hardware, software, other logic, or combination of the preceding for servicing the requests from one or more other computers. The term “computer system” may generally refer to a system including one or more server computers coupled to one or more databases.DETAILED DESCRIPTION
[0037] Embodiments described herein are directed to methods and systems of efficient detection of potentially invalid activity in relation to access requests.
[0038] Existing methods of analyzing access requests as valid or invalid requests consider certain data values connected with the access request to determine a risk level associated with the access request. Often, the more data values associated with an access request considered, the more accurate the risk level determination becomes. However, this comes at the expense of speed and performance as considering additional data values is resource intensive and computationally expensive. A method for an efficient detection mechanism that can handle high-dimensional inputs efficiently is desirable.
[0039] The techniques disclosed herein can consider, in addition to historical access request data, high-dimensional input when analyzing riskiness of new access requests. For example, an indicator that an access request may be invalid is if the IP address associated with the access request has been used in connection with multiple unique email addresses. By779088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 extending this concept and observing how many unique data points this particular IP address is linked to for a variety' of fields (transaction device ID. shipping address, etc.), the system can assign a more accurate risk evaluation score. This risk evaluation score may determine whether or not the access request to obtain a resource is granted.
[0040] Embodiments can provide several advantages over conventional methods of determining validity' of an access request. The techniques described herein provide a means of representing complex, high-dimensional data in an organized, condensed, and integrated manner, thereby reducing scalability issues tied to memory and processing requirements for handling such large data sets. Therefore, a detection mechanism that requires a requisite number of dimensions to reach a threshold level of accuracy need not compromise speed and performance to obtain such accuracy since the dimensions are extracted and represented as a singleton, or "‘thread’;
[0041] Before the present disclosure is described in greater detail, it is to be understood that this disclosure is not limited to particular embodiments described, as such may vary. It is also to be understood that the terminology' used herein is for the purpose of describing particular embodiments only, and is not intended to be limiting, since the scope of the present disclosure will be limited only the appended claims.I. AUTHENTICATION FOR ACCESSING A PROTECTED RESOURCE
[0042] Generally, access requests for a computer resource or account (e.g., transactions over the Internet) go through a detection system to determine whether the transaction is valid or invalid. Thus, a resource security system may receive requests to access a resource. The resource security system may include an access server for determining an outcome for the access request based on access rules. An exemplary' resource security system is described in further detail below.
[0043] FIG. 1 shows a resource security system 100 for authorizing access to resources, in accordance with some embodiments. The resource security system 100 may be used to provide authorized users (e.g., via authentication) access to a resource while denying access to unauthorized users. In addition, the resource security' system 100 may be used to deny invalid access requests that appear to be legitimate access requests of authorized users. The resource security system 100 may implement access rules to identify invalid access requests879088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 based on parameters of the access request. Such parameter may correspond to fields (nodes) of a data structure that is used to distinguish invalid access requests from valid access requests.
[0044] The resource security system 100 includes a resource computer 110. The resource computer 110 may control access to a physical resource 118, such as a building or a lockbox, or an electronic resource 116, such as a local computer account, digital files or documents, a network database, an email inbox, a payment account, or a website login. In some embodiments, the resource computer may be a webserver, an email server, or a server of an account issuer. The resource computer 110 may receive an access request from a user 140 via a user device 150 (e.g., a computer or a mobile phone) of the user 140. The resource computer 110 may also receive the access request from the user 140 via a request computer 170 coupled with an access device 160 (e.g.. a keypad or a terminal). In some embodiments, the request computer 170 may be a resource provider. For example, the request computer 170 and the resource computer 110 may be the same, where the access request from the user 140 is generated directly at the resource computer 110.
[0045] The access device 160 and the user device 150 may include a user input interface such as a keypad, a keyboard, a finger print reader, a retina scanner, any other type of biometric reader, a magnetic stripe reader, a chip card reader, a radio frequency identification reader, or a wireless or contactless communication interface, for example. The user 140 may input authentication information into the access device 160 or the user device 150 to access the resource. Authentication information may also be provided by the access device 160 and / or the user device 150. The authentication information may include, for example, one or more data elements of a user name, an account number, a token, a password, a personal identification number, a signature, a digital certificate, an email address, a phone number, a physical address, and a network address. The data elements may be labeled as corresponding to a particular field, e.g., that a particular data element is an email address. In response to receiving authentication information input by the user 140, the user device 150 or the request computer 170 may send an access request, including authentication information, to the resource computer 110 along with one or more parameters of the access request.
[0046] In one example, the user 140 may enter one or more of an account number, a personal identification number, and password into the access device 160, to request access to979088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 a physical resource (e.g., to open a locked security door in order to access a building or a lockbox) and the request computer 170 may generate and send an access request to the resource computer 110 to request access to the resource. In another example, the user 140 may operate the user device 150 to request that the resource computer 110 provide access to the electronic resource 116 (e.g., a website or a file) that is hosted by the resource computer 110. In another example, the user device 150 may send an access request (e.g., an email) to the resource computer 110 (e.g., an email server) in order to provide data to the electronic resource 116 (e.g., deliver the email to an inbox). In another example, the user 140 may provide an account number and / or a personal identification number to an access device 160 in order to request access to a resource (e.g., a payment account) for conducting a transaction.
[0047] In some embodiments, the resource computer 110 may verify the authentication information of the access request based on information stored at the request computer 170. In other embodiments, the request computer 170 may verify the authentication information of the access request based on information stored at the resource computer 110.
[0048] The resource computer 110 may receive the request substantially in real-time (e.g., account for delays computer processing and electronic communication). Once the access request is received, the resource computer 110 may determine parameters of the access request. In some embodiments, the parameters may be provided by the user device 150 or the request computer 170. For example, the parameters may include one or more of: a time that the access request was received, a day of the week that the access request was received, the source-location of the access request, the amount of resources requested, an identifier of the resource being request, an identifier of the user 140, the access device 160, the user device 150, the request computer 170, a location of the user 140, the access device 160, the user device 150, the request computer 170, an indication of when, where, or how the access request is received by the resource computer 110, an indication of when, where, or how the access request is sent by the user 140 or the user device 150, an indication of the requested use of the electronic resource 116 or the physical resource 1 18, and an indication of the type, status, amount, or form of the resource being requested. In other embodiments, the request computer 170 or the access server 120 may determine the parameters of the access request.
[0049] The resource computer 110 or the request computer 170 may send the parameters of the access request to the access server 120 in order to determine whether the access request is1079088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 invalid. The access sen' er 120 may store one or more access rules 122 for identifying an invalid access request. Each of the access rules 122 may include one or more conditions corresponding to one or more parameters of the access request. The access server 120 may determine an access request outcome indicating whether the access request should be accepted (e.g., access to the resource granted), rejected (e.g., access to the resource denied), or reviewed by comparing the access rules 122 to the parameters of the access request as further described below. In some embodiments, instead of determining an access request outcome, the access server 120 may determine an evaluation score based on outcomes of the access rules. The evaluation score may indicate the risk or likelihood of the access request being invalid. If the evaluation score indicates that the access request is likely to be invalid, then the access server 120 may reject the access request.
[0050] The access server 120 may send the indication of the access request outcome to the resource computer 110 (e.g., accept, reject, review, accept and review, or reject and review). In some embodiments, the access server 120 may send the evaluation score to the resource computer 110 instead. The resource computer 110 may then grant or deny access to the resource based on the indication of the access request outcome or based on the evaluation score. The resource computer 110 may also initiate a review process for the access request.
[0051] In some embodiments, the access server 120 may be remotely accessed by an administrator for configuration. The access server 120 may store data in a secure environment and implement user privileges and user role management for accessing different types of stored data. For example, user privileges may be set to enable users to perform one or more of the following operations: view logs of received access request, view logs of access request outcomes, enable or disable the execution of the access rules 122, update or modify' the access rules 122, change certain access request outcomes. Different privileges may be set for different users.
[0052] The resource computer 110 may store access request information for each access requests that it receives. The access request information may include authentication information and / or the parameters of each of the access requests. The access request information may also include an indication of the access request outcome for the access request, e.g., whether access request was actually fraudulent or not. The resource computer 110 may also store validity information corresponding to each access request. The validity1179088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 information for an access request may be initially based on its access request outcome. The validity information may be updated based on whether the access request is reported to be fraudulent. In some embodiments, the access server 120 or the request computer 170 may¬ store the access request information and the validity information.II. DATA ELEMENTS AS DETECTORS
[0053] An access request may contain one or more data elements associated with the access request. It is desirable to represent the interrelationship of data elements by using a segment when validating an access request. A segment may be a representation of the relationship between a particular detector chosen from amongst the selected detectors and the other detectors associated with the access request. For instance, a segment may correspond to the number of values of other detectors used historically for a specific detector. A segment may also represent whether a detector has the same value as other detectors within a set of detectors. A thread is the set of segments associated with an access request. Furthermore, a thread can represent the relationship between the detectors for a single access request. In some instances, a thread can represent the relationship between a set of historical access requests in relation to the current access request.
[0054] Accordingly, embodiments of the disclosure can use one or more detectors in determining the validity7of one or more data elements (e.g., extracted from user inputted authentication information) related to an access request. The validity- of a data element can be used to determine a validity- of an access request including the data element. The contributions of data elements (negatively or positively) toward validity7or suggesting an attacker can be used to make a final determination for whether to authenticate an access request as part of providing access to the resource based on the access request outcome.
[0055] A detector can provide a validity contribution using historical access requests and the corresponding results that included the data element(s) of that detector. In this manner, a detector can extract the data element(s) assigned to that detector from an access request. For example, a particular detector can be assigned to the email in the access request. The detector can then use historical access requests that included the given email of an access request to determine a validity- contribution, where such historical access requests can be for a particular resource or across a set of resources. The more valid results from the historical access1279088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 requests generally provides a more positive validity7contribution. The more invalid results (e.g., identified as an attacker, usually after the fact), the more negative the validity contribution would be. Various types of historical access result values can affect the validity contribution in different ways, e.g., with different result values having different weights.
[0056] An individual detector may be configured to evaluate a single data element (e.g., an account number). Another detector may further be configured to evaluate a combination of at least two or more data elements (e.g., an account number and an email). In some embodiments, a set of detectors is defined, where at least a subset of the detectors validate a single data element and another subset of the detectors validate combinations of two or more data elements. A data element may be evaluated individually and in combination with other data elements by more than one detector. For example, a set of detectors may include (but is not limited to) a detector for an account number and another detector for an account number and an email.
[0057] Each of the set of detectors may retrieve a vector of historical access result data associated with its corresponding one or more data elements for a period of time (e.g., one or more years). The vector may be retrieved from a data structure stored within a database. The vector of historical access result data can comprise a set of counts for past access results associated with the corresponding one or more data elements. That is, different types of access results (e.g., accept, reject, fraud, and review) can each have their own count. Further, different subtypes can have their own count, e.g., different types of acceptances can contribute different amounts to an overall acceptance count.
[0058] For instance, the set of counts can include a count of accepted access requests and rejected access requests. For example, a detector for a user email may retrieve a vector that includes the number of times that access requests associated with the user email have been accepted or rej ected. The set of counts may further specify a count of access requests for which a determination could not be made and a count of access requests that were initially accepted but later determined to be fraudulent.
[0059] In some embodiments, the historical access result data may be further differentiated into a set of predetermined time interv als (e.g., weekly or monthly) over the period of time. Furthermore, each vector of historical access result data can be identified by a resource identifier, such that when retrieving the vector, the resource identifier associated with the 1379088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 vector is compared to the resource identifier provided within the access request. In other implementations, the vector of historical access result data can be an aggregate for all resources, and thus agnostic to the particular resource identified in a current access request.
[0060] After retrieving the vector of historical access result data for each of the set of detectors, a detector value for each detector can be generated based upon the retrieved vector of historical access result data. Each detector value may predict a probability7or likelihood that the corresponding one or more data elements are valid and not associated with a malicious access request, e.g., as a validity contribution. A vector of detector values is generated, where each of the detector values is generated from one of the set of detectors. The vector of detector values can be aggregated into an evaluation score, where the evaluation score determines either a positive classification or a negative classification for the access request. In some embodiments, at least one of the conditions for an access rule defined by the access server for authenticating an access request is based upon the determined classification. For example, an access rule may include a condition to reject any access request that has been given a negative classification or a certain level of a negative classification, e.g., for a highly negative classification.
[0061] The set of detectors may further be categorized into single provider detectors and cross provider detectors. A single provider detector is used with a distribution determined for only a particular resource provider (or request computer) associated with the current access request in order to provide a first classification. A cross provider detector can be used in combination with a distribution determined across multiple resource providers in order to provide a second classification. Therefore, an access request may be given a first classification based upon a set of single provider detectors and a second classification based upon a set of cross provider detectors. In some instances, the first and second classifications for the current access request may not be the same.III. PERFORMANCE AT HIGHER DIMENSIONS
[0062] Embodiments disclosed herein provide several benefits with regards to detection of unauthorized activity. Access requests may be incorrectly flagged as invalid, therefore requiring entities to dedicate resources to review and correct these access requests. For entities that process millions of access requests a day, this translates into several hundreds of1479088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 thousands of access requests that may be incorrectly flagged. One such benefit disclosed herein is, that by increasing the number of dimensions (e.g., the number of fields used) considered by the detection mechanism, the accuracy of the detection mechanism increases. Even by just increasing the number of dimensions considered from just one dimension to two dimensions can result in a 5-fold decrease in false positives as further described below.
[0063] FIG. 2A depicts a graph of the relationship between accuracy and number of dimensions utilized, using the embodiments disclosed herein. The number of dimensions considered when determining whether an access requests is valid or not is represented along the x-axis and the false positives ratio (FP_Ratio) is represented along the y-axis. The FP_Ratio is calculated by taking the total number of access requests incorrectly marked as invalid divided by the total number of access requests marked as actually invalid or incorrectly marked as invalid. This data shows that, with each increase in dimension, the ratio of false positives decreases. Thus, using higher dimensions increases accuracy.
[0064] FIG. 2B depicts a graph of the relationship between accuracy and number of dimensions utilized, using the embodiments disclosed herein. The number of dimensions is represented along the x-axis and the ‘‘risk index"’ is represented along the y-axis. Here, the risk index is calculated to reflect the proportion of invalid access requests to the total number of all access requests flagged as invalid or incorrectly marked as invalid. It is calculated by dividing the total number of valid access requests by the total number of invalid access requests (either determined to be invalid or incorrectly marked as invalid). Therefore, a higher risk index calculation corresponds to a higher number of valid access requests in comparison to the total number of invalid access requests. As illustrated by the graph, when the number of dimensions increase, the risk index also increases. Thus, using higher dimensions translates into higher accuracy when determining the validity of an access request.
[0065] The foregoing illustrates the benefit of increasing the number of dimensions considered when analyzing an access request, according to some embodiments described herein. However, increasing the number of dimensions increases the computational resources required since the number of fields increases, along with all of the possible values for each field. For example, for each account there may be 2-10 different values for each field, so five fields can result in 105combinations, and these combinations can occur for millions of1579088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 accounts. Embodiments of the present disclosure can reduce the required computational resources relative to tracking all combinations, particular when there are at least 10, 20, 50.100, 1,000. 2.000, 5,000. 10.000. 100,000 different combinations of values for fields used to track access requests and determine whether they should be authorized as being valid access requests. And such embodiments can involve at 10,000, 100,000, and one million accounts, the benefits can be particularly advantageous.IV. DETECTION SYSTEM AND THREAD GENERATION
[0066] Access requests to a resource (e.g., a building, a computer, a website, or an electronic account) go through a detection system whereupon the access request is valid or invalid based upon a determination by the detection system. An exemplary detection mechanism is described in further detail below.
[0067] As mentioned above, a cluster thread is generated from a set of one or more segment threads. A cluster thread comprises the set of one or more segment threads generated for a single access request. In situations where only one segment thread is generated, the cluster thread and segment thread are equal. A segment thread corresponds to a set of category labels indication the relationship between detectors for a single access request. A cluster thread can also represent the relationship between a set of historical access requests and the current access request. The validity of an access request can be determined by analyzing its cluster thread.A. Creating historical access result index
[0068] FIG. 3 A shows a detection system 300 for generating a historical access result index. The detection mechanism 300 may be hosted on a server or other suitable means. The detection mechanism 300 may include a historical access result database 330 for the storage of data structures comprising historical access result data. The historical access result database 330 may be queried to retrieve information related to the historical access result data. For instance, as will be further described below in section V, the count of unique detector values associated with a particular detector value may be determined by querying the historical access result database 330. The historical access result database 330 may further comprise a routing table where each of the stored data structures is mapped to a resource identifier 301 A.1679088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01
[0069] In step SI, the detection system 300 may receive (e.g., retrieved) a historical access request 301 stored in the historical access result database 330. The historical access request 301 may comprise a resource identifier 301A and a set of data elements 301B. Many historical access requests can be received for various resources (each having a different resource identifier), and many access requests can be for a same resource. These historical access requests are associated with known outcomes (e.g., valid or not valid) as determined over time, e.g., from computer resources 320. These outcomes are shown as outcome labels 308, which may be determined after invalid activity has been detected in the systems later (e.g., after a cyberattack). In other implementations, such outcome labels 308 can be stored in historical access result database 330.
[0070] In step S2, the historical access request 301 is transmitted to the detector extractor 302. The detector extractor 302 can extract, from the set of data elements 301B from the historical access request 301, a subset of data elements. Extraction of the subset of data elements is based on a predetermined set of fields determined to be the detectors. The predetermined set of fields can contain at least three unique fields (e.g., email, IP address, billing address, currency country, IP country, etc.). Detector extractor 302 may communicate with the resource provider (not shown) to receive a set of at least three fields that specify a subset of the set of a data elements that should be used as detectors w hen generating the historical access request index. The set of at least three fields can be an input transmitted to the detection mechanism 300 from a client device or other suitable device (not shown). The set of fields can be used for tracking invalid requests and to determine rules for identifying future access requests as invalid or valid.
[0071] The detector extractor 302 may receive more than one set of fields determined to be detectors if the detection mechanism 300 determines an access request may be compared against multiple different types of cluster threads. The types of cluster threads may differ if the cluster threads were generated from different sets of fields and / or different seed detectors. For example, the first set of fields may list email, IP address, and billing address as the subset of data elements that should be used as detectors. A second set of fields may list billing country, shipping country, resource country (country' where the resource is located, such as a physical building), currency country, actual IP country’ (country the transaction is originating from), and proxy IP country (country7presented during the transaction) as the subset of data1779088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 elements that should be used as detectors. Therefore, the cluster thread generated from first set of fields may be compared against cluster threads generated using the same fields as the first set of fields, and the cluster thread generated from the second set of fields may be compared against cluster threads generated using the same fields as the second set of fields. The detector extractor 302 may extract data elements corresponding to the fields listed in the first set of fields. Detector extractor 302 may then extract data elements corresponding to the fields listed in the second set of fields. The detector extractor 302 may then transmit data elements corresponding to the first and second set of fields in accordance with S3, described in greater detail herein.
[0072] In step S3, the extracted subset of the set of data elements selected to be detectors is transmitted to the segment module 304. The subset of the set of data elements may be transmitted in the form of a vector or other suitable data structure. Furthermore, the vector may contain the resource identifier 301 A associated with the historical access request 301 such that the generated segment threads may be appropriately categorized as described in greater detail below. The segment module 304 may comprise subset segment modules 304A, 304B. and 304C. Each subset module of segment module 304 may be assigned to process a set of data elements extracted by the detector extractor as described in step S2. In some embodiments, a subset segment module may handle generating segment threads for all the sets of data elements extracted by the detector extractor as described in step S2.
[0073] As described above, a segment thread is the collection of category labels assigned to the detectors in a set of detectors, and the category labels may correspond to a predetermined threshold value (e.g., high, medium, low) or other suitable metric. In some embodiments, the category labels may correspond to a predetermined value indicating a pairwise comparison result (e.g., same, different, etc.).
[0074] After receiving a set of data elements, the segment module can generate a segment thread for each detector identified as the seed detector amongst the set of detectors. The seed detector indicates which detector value will be used as a reference value when generating the segment thread, described in further detail herein. The indicator may be communicated to the segment module 304 from the resource provider, from a client device, or other suitable device. The detector's position within the set of detectors may also indicate if the detector is a seed detector. If no seed detector is indicated, the segment module 304 may select the first1879088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 detector in the set of detectors to be the seed detector. More than one seed detector may be indicated in the set of detectors and each detector may be indicated to be seed detector. In such instances, multiple segment threads may be generated.
[0075] In step 4, the segment module 304 may query the historical access result database 330 while other embodiments may proceed without such a query. For example, the segment module 304 may access the historical access result database 330 to determine, for each detector in the set of detectors, the count of unique detector values associated with the seed detector value. Stated alternatively, segment module 304 may determine how many unique instances a particular detector value been used in connection with another detector value. The count of unique detector values for each detector may be stored in a vector corresponding to the seed detector. Each count may be stored in the vector such that the count's index position corresponds to its respective detector’s index position within the vector transmitted in step S3. The order in which the counts are stored in the vector may correspond to the order in which the detectors were stored in the vector transmitted in step S3. Furthermore, the resource identifier 301 A may also be stored in the vector at a suitable index position. For instance, a set of detectors may contain email, IP address, and billing address. The set of detectors may be transmitted as vector (i). If the seed detector is email, then the segment module may determine the seed detector value for email has been associated with 5 unique IP address values. Moving to the next detector, billing address, the segment module may then determine the seed detector value for email has been associated with 8 unique billing address values. Therefore, the resulting vector generated for the set of fields comprising email, IP address, and billing address may resemble vector (ii).#0000 Email IP Address Billing Address#0000 * 5 8Here, (*) is used to denote the seed detector and #0000 is an example resource ID.Furthermore, each count in vector (ii) is stored in the same index position as its corresponding detector in vector (i).1979088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01
[0076] The segment module 304 may assign a category label to each detector value in the vector to generate a segment thread. Each category label is defined by a predetermined threshold metric or result. For example, the category labels may be “low”, “medium”, and “high” and each category label may be defined by a range of values. As another example, the category7label may correspond to a comparison result such as: “same”, “different”, “empty ”, or “not applicable”. For example, a segment thread generated from vector (ii) may resemble the string depicted in (iii).(in) * L MWhere “L” denotes the category label “low” and “M” denotes the category label “medium” based on some predetermined threshold and the detector counts in vector (ii).
[0077] At step S5, once the segment module 304 has generated the segment thread(s) associated with the detectors, the segment module 304 can combine the one or more segment threads into a cluster thread and transmit a cluster thread generated from the set of segment threads to the thread metrics database 306. If the cluster thread has one segment thread, then the segment thread is equivalent to the cluster thread. The thread metrics database 306 stores the cluster thread in a historical access result index, which may be organized as a table. The table rows may correspond to cluster threads and the table columns may represent metrics and values associated with each cluster thread. Metrics may include those such as percentage of access requests corresponding to that particular cluster thread that are reported as a fraudulent, false positive, chargeback, or authorized request.
[0078] The thread metrics database 306 may update the values of the historical access request index based on outcome labels 308 received from the resource computers 320. The resource computers 320 can determine and store the outcome of each historical access result, the outcome indicating whether the historical access result was valid or invalid. Thread metrics database 306 can identify the cluster thread corresponding to the outcome labels 308 and update the metrics accordingly. As described above, each row of the table may correspond to a unique cluster thread. The columns of the table may correspond to fields that contain statistical information characterizing the validity' of each respective cluster thread. For instance, for a given cluster thread, the first column may comprise a field storing count values representing the number of instances in which the corresponding cluster thread was determined to be valid. The second column may comprise a field storing count values 2079088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 representing the number of instances in which the corresponding cluster thread was determined to be invalid. The third and fourth columns may comprise fields representing the percentage of the corresponding cluster thread determined to be valid and invalid, respectively. One or more other validity or invalidity metrics can be used additionally or alternatively.B. Using historical access result index
[0079] FIG. 3B shows a detection mechanism 300 for authorizing access to resources for an incoming current access request, in accordance to some embodiments. The detection mechanism 300 may be used to provide authorized users (e.g., via authentication) access to a resource while denying access to unauthorized users. The detection mechanism 300 may generate a data vector from high-dimensional data associated with an access request, thereby creating a representation of the access request as a singleton. Handling a singleton as opposed to high-dimensional data is less computationally expensive and may increase the speed of analyzing access requests.
[0080] A current access request 350 transmitted to the detection system 300 may include a resource identifier 350A and data elements 350B. The current access request 350 may originate from a client device (not shown) or other suitable device. The current access request may comprise a resource identifier 350A and data elements 350B, as described herein.
[0081] In step SI, the detector extractor 302 receives the current access request 301. As described above, a predetermined set of fields determined to be the detectors is communicated to the detector extractor 302. The detector extractor 302 extracts, from the current access request 301, a subset of data elements from the set of data elements 301B based on the set of fields. The set of fields may indicate the data elements determined to be detectors and the detector identified to be the seed detector. The extracted subset of the set of data elements selected to be detectors as well as the resource identifier 301 A associated with the current access request 301 may be stored as a vector.
[0082] In step S2, a segment module 304 receives the vector containing the subset of data elements selected to be detectors and the resource identifier 301A. Segment module 304 may generate a segment thread associated with the data value identified to be the seed detector for the current access request 301 according to the methods disclosed herein.2179088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01
[0083] In step S3, a cluster thread is generated from the one or more segment threads generated for the current access request 301 and is received by the historical access result index module 310. The historical access result index module may determine whether to authorize the current access request 301. Furthermore, the historical access result index module 310 may comprise a comparison module 310A and access request determination module 310B. After receiving the cluster thread, the historical access result index module 310 may request and receive a historical access result index, which may be in the form of a table of cluster thread metrics from the thread metrics database 306.
[0084] The comparison module 310A, utilizing the cluster thread as an identifier, can match the cluster thread corresponding to the current access request to a cluster thread contained in the historical access result index. Based on the metrics associated with the matching cluster thread, the access request determination module 310B may then determine the current access request 301 is authorized or unauthorized. Access request determination module 310B may make the determination by assessing if the metrics and / or values associated with the cluster thread exceed a predetermined risk tolerance threshold and / or violate one or more rules. For example, if the cluster thread is associated with a chargeback percentage exceeding a particular threshold, the access request determination module may deem the current access request as unauthorized. As another example, a rule may stipulate an access request be denied if its corresponding cluster thread matches a cluster thread with a history of invalid requests exceeding a threshold percentage.
[0085] In step S5, the determination from access request determination module 310B is transmitted to external entity 360. The external entity 360 may control access to the requested resource. Based on the determination received, the external entity 360 may then deny access to specified resource identified by resource identifier 350A from the current access request. In addition, the external entity 360 may store the determination or transmit the determination to different external device for further analysis.V. CROSS COMBINATION MORPHING DETECTOR
[0086] As described above, accuracy in detecting an invalid access request increases the more dimensions a detection mechanism considers. Each access request may be associated with a multitude of fields containing data that may be relevant in evaluating the access2279088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 request's risk potential. The challenge posed by utilizing the numerous field data associated with an access request lies in efficiently processing such quantities of data without compromising computer system performance.
[0087] The cross combination morphing detector builds upon the methods and techniques discussed above. As will be discussed in further detail below, the cross combination morphing detector is used to generate segments for historical access requests based on a predetermined set of detectors. Each segment may represent the relationship of the seed detector and other detectors in the set of detectors. Specifically, each segment represents how many unique detector values have been used in connection to the seed detector for each detector in the set of detectors. Multiple segment threads may be generated for each access request and the multiple segment threads may be aggregated to constitute a cluster of threads associated with the access request. By representing high-dimensional input as threads, complex relationships and patterns can be more efficiently processed and analyzed, enabling streamlined computations and improved scalability of access request validation.A. Generating segment threads and cluster threads
[0088] A sequence of detectors may be formed to generate a cluster thread to represent the high-dimensional input as a single string, thereby reducing the computational burden of processing such information. A cluster thread is generated from a set of segment threads and each segment thread is generated from a set of category labels representing the intrarelationship of the selected detectors. In some instance, only one segment thread may be generated and so the cluster thread and segment thread are equal. One technique for generating a cluster thread representing the relationship of detectors used in connection with other detectors is described below.
[0089] FIG. 4A illustrates an example segment thread generated for a particular current access request (e.g., current access request 350). in accordance to some embodiments. The system may receive a current access request and data elements (e.g., data elements 350B) corresponding to fields associated with the current access request . The access request and its corresponding data elements may be stored in, and accessible from, a database (e.g., database 306) or other suitable means. At least three detectors, such as 402A-414A, are selected from the set of fields associated with the access request based upon predetermined criteria. The2379088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 criteria for selecting a set of detectors may depend on the observed accuracy performance of such fields selected to be detectors. The selected detectors may be stored in any suitable data structure such as a vector and the position of the detector within the vector may indicate its importance in comparison to the other detectors. Additionally, the detector may be referenced by its position in the vector (e.g., 1, 2, 3, etc.) or by an abbreviation assigned to that vector (e.g., “E” for email, “B” for billing address, “A” for IP address, etc.).
[0090] The process further includes selecting one detector amongst the set of detectors 402A-414A to be the seed detector. The seed detector, or simply the seed, serves as the reference value from which the category labels are determined for each of the other detectors in the set of detectors. For each non-seed detector, the total number of unique values belonging to that non-seed detector that have been observed to be associated with the seed detector’s value is determined, as illustrated by 416A-428A. In other words, how many unique detector values, for a given detector, has the seed value been associated with from a given set of access results. The numeric value representing the number of unique detector values the seed value has been associated with determines the category7label assigned to that detector. Category labels, as illustrated by 430A-442A, may represent a score (e.g.. high, medium, low) or other suitable metric. Each category label may correspond to a threshold value so that each detector is assigned a category' label based on the numeric value determination described above. For example, if a seed value was determined to have been used in connection with 3 unique email addresses and 10 billing addresses, then the category label assigned to the email detector may be “low” and the category label assigned to the billing address detector may be “medium”. The seed detector is not given a label and may be represented by a symbol (e.g., “*”), as illustrated by 420 A.
[0091] Once each non-seed detector is assigned a category7label, the category7labels are then concatenated to form a string of category labels, or “segment thread”. The position of the category label in the string corresponds to the position its respective detector in the vector. The symbol denoting the seed detector is also represented in the segment thread and its position in the segment thread also corresponds to the seed detector’s position in the vector. For example, for a set of detectors of size 3 and where the seed detector is in the first position within the vector, the resulting segment thread may be:* LM2479088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01
[0092] The same process as described above may be repeated as necessary7for some or all of the other detectors in the set of detectors. For example, after generating the segment thread for a selected seed detector, a different detector in the set of detectors may be chosen to be the seed detector. As described above, a segment thread is then generated for the second seed detector. The set of segment thread(s) generated for one or more detectors in a set of detectors may be concatenated together to form a cluster thread. Each segment thread in the concatenated string of segment threads may be separated by a dash or other suitable marker. For example, a dashmay be appended to the end of the segment thread generated for detector 1 and the segment thread for detector 2 may be appended after the dash. The set of segment threads may be used to generate a cluster thread. A cluster thread can be the same as the segment thread if only one segment thread is generated.B. Generating and using a historical access result index
[0093] FIG. 4B is an example historical access result index 400B generated from cluster threads. The historical access result index 400B may be generated using the methods and embodiments disclosed herein. Each row may correspond to a cluster thread generated for a historical access request of a set of historical access request (e.g., historical access request 301). Each column may contain a metric or statistical information characterizing the validity of the respective cluster thread. Metrics may include those such as percentage of access requests corresponding to that particular detector thread that are reported as a fraudulent, false positive, chargeback, or authorized request. The metrics and statistical information corresponding to each detector thread may7be updated when an outcome label assigned to the access request corresponding to the detector thread is received, as described herein.
[0094] The resulting geographical detector thread generated for a current access request may be used as an identifier to identify a matching cluster thread (e.g., 420B-430B) in the cluster thread column 402B of the historical access result index 400B. Once a matching thread in the cluster thread column 402B is found, the corresponding metrics from columns 404B-416B may be used to determine whether or not to grant access to the resource identified by the current access request. Access request may be determined by assessing if the metrics and / or values associated with the cluster thread 420B-430B exceeds a predetermined risk tolerance threshold and / or violate one or more rules. For example, if the cluster thread is2579088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 associated with a chargeback percentage exceeding a particular threshold, the access request determination module may deem the current access request as unauthorized. As another example, a rule may stipulate an access request be denied if its corresponding cluster thread matches a cluster thread in the historical access result index 400B with a history of invalid requests exceeding a threshold percentage.C. Example
[0095] A cluster thread may be generated for every access request. For a given access request, at least three unique detectors are selected from the set of fields associated with the access request. For example, the set of fields associated with an access request stemming from an online transaction may contain 10 different fields: billing address, account number, email, purchase device. IP address, phone number, shipping address, customer name, date of transaction, and merchant name. From this list of 10 fields, the fields chosen to generate the set of 7 detectors may contain: billing address, account number, email, purchase device, IP address, phone number, and shipping address (e.g., 402A-414A).
[0096] The detectors 402A-414A may be stored in the resulting vector from most important to least important detector and subsequently referred to by its assigned index number.References made to a particular detector may be done so by referencing its assigned index number.
[0097] A detector can then be selected from the set of at least three unique detectors to be the seed detector. In this example, the seed detector 406A is the selected seed detector. Each detector in the set of detectors, except the seed detector, is analyzed to determine the total count of unique detector values belonging to that detector in association with the seed detector. Each detector is then assigned a category' label (e.g., L = low, M = medium, H = high, etc.) based on where the count falls in a predetermined range. The selected detectors may, but not necessarily, be ordered from greatest to least importance for predicting a correct outcome value and the assigned index number may indicate the detector’s importance ranking amongst the set of detectors. A detector’s importance may be determined by analyzing the detector’s discriminatory power using methods such as Receiver Operating Characteristic (ROC) curves, Area Under the Curve (AUC), Youden Index, and / or any other suitable means.2679088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01
[0098] Continuing from the abovementioned example, the selected detectors are: billing address corresponding to detector 402A, account number corresponding to detector 404A, email corresponding to seed detector 406 A, purchase device corresponding to detector 408A. IP address corresponding to detector 410A, phone number corresponding to detector 412A, and shipping address corresponding to detector 414A. Email may be the detector selected to be the seed detector 406A. From a set of access requests, the total unique detector values 416A-428A for each detector in the set of detectors with respect to the seed detector is tabulated. For illustrative purposes, the email address value (e.g., johnny.appleseed@email.com) for seed detector email has been used in connection with 1 billing addresses, 11 account numbers, 5 purchase devices, 2 IP addresses, 1 phone number, and 3 shipping addresses. A predetermined range may be, as an example:Low: 0 < total detector count < 3Medium: 3 < total detector count < 10High: 10 < total detector countTherefore, a resulting thread may resemble segment thread 440A.
[0099] In this example, only one detector was selected to be the seed detector 406 A.Therefore, the resulting segment thread 440A is the cluster thread for the access request associated with the detectors 402A-414A.D. Method
[0100] FIG. 5 depicts a flow chart of a method 500 for securing access to resources by determining the validity of an access request in accordance with some embodiments of the disclosure. The method may be performed by the detection mechanism 300 of FIG. 3 A or other suitable means. The detection mechanism can validate incoming access requests in realtime.
[0101] In step 502, a set of access requests (e.g.. historical access request 301)may be received over a network from a plurality of devices. Each of the access requests may be associated with a resource identifier (e g., resource identifier 301A) for a respective resource and a plurality7of detector values corresponding to at least three fields (e.g., email, billing address, IP address, etc.). The set of access requests may include at least 500,000 access requests.2779088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01
[0102] In step 504, the access result values for the set of access requests may be stored in a database (e.g., database 306) or other suitable means and accessible to the detection mechanism 300. The access result value may provide a measure of approval in granting valid access to the respective resource. In other words, the access result values may indicate past approved or rejected access requests. The access result value may be transmitted by a suitable device (e.g., resource computer 320). The access result values may be transmitted as a Boolean, text string, or other suitable indicator (e.g.. outcome labels 308) that indicates, for example, if the access result was valid or invalid.
[0103] In step 506, a first field of the at least three fields associated with the set of access requests is determined to be the first seed field. The detector values associated with the at least three fields associated with an access request may be stored in a vector. The resource identifier associated with the access request may also be stored in the vector. The first seed field is associated with a particular detector value that is to be used in generating category labels for the other fields of the at least three fields. The at least three fields associated with the set of access requests may be determined by a subsystem of the detection mechanism (e.g., detector extractor 302) or other suitable means.
[0104] The first seed field can be selected based on a ranking of fields where the fields may be ranked for its predictive capabilities in determining the validity of an access request. The predictive capabilities of a particular field may be determined through observation of the field’s performance in predicting the validity of an access request. For example, testing may be conducted on a set of historical access results with known access outcomes to determine the accuracy of each field in predicting the validity' in of an access request.
[0105] The at least three fields may be selected from: billing address, account number, email, purchase device, IP address, phone number, and shipping address.
[0106] In step 508, a category' label is determined for other fields of the at least three fields. The category label can indicate an amount of different detector values in access requests that include the particular detector value of the first seed field. The category labels generated may¬ be stored in a vector and each category' label’s index position may correspond to its respective detector’s index position. A subsystem of the detection mechanism (e.g., segment module 304) or other suitable means may handle the generation of category labels and the2879088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 storage of said category labels in a vector. The category labels may be predetermined values that may indicate a range, type, result, or other suitable metric.
[0107] The category label may correspond to a range of values, the range being specified by an upper and lower boundary values. For example, if the set of category labels are “low”, “medium”, and “high” and correspond to the number of unique detector values detected, then each category label may correspond to a range of counts of unique detector values.Continuing with this example, “low” may correspond to 0-5 detector values counted, “medium” may correspond to 6-10 detector values counted, and “high” may correspond to 11+ detector values counted. In addition, the category labels associated with an access request may be stored in a vector. The vector may also contain the resource identifier associated with the access request.
[0108] In step 510, a thread of category labels may be generated for the particular detector value, thereby generating a set of segment threads. The set of segment threads may generate a cluster thread. For example, the segment module may generate the cluster thread by storing the set of segment threads as vectors, and each segment thread may correspond to its ow n vector. Each category label associated with a segment thread may be stored in a unique index position in the vector. The category label’s index position may indicate the respective detector the category label is assigned to.
[0109] The cluster threads may be transmitted to a database w here it may' be sorted into a searchable table (e.g., historical access result index 400B). The historical access result index may comprise a plurality of rows, each row corresponding to a cluster thread of the set of cluster threads (e.g., cluster thread 420B-420B) and a plurality’ of columns, each column representing an access metric associated with the corresponding cluster thread (e.g., 404B-416B).
[0110] In step 512, the detection mechanism may determine one or more access metrics from the access result values received. The access result values may be received, by the thread metrics database (e.g.. database 306), from a device (e.g., resource computer 320) or other suitable means. The access result value may correspond to a cluster thread and used to update the database 306 storing metrics on the cluster threads. The access result value may be received from a resource computer 320 or other suitable means. The access result value may¬ be stored in a column of a table (e.g., cluster thread column 402B) associated with the cluster 2979088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 threads. The access metrics may be determined by calculating metrics, such as percentage of invalid access results corresponding to the access request, using the access result values. The database may be updated with the determined access metrics.
[0111] The access metric may be updated periodically based on new access values received. Based on the new access value received, the access metrics may be updated to reflect the new access value. For example, an access metric measuring the percentage of a particular access request found to be invalid may be recalculated if the new access value received indicates the access request was recently found to be valid.
[0112] In step 514, the detection mechanism may receive, over a network from a plurality of devices (e.g., a user device, an access device, request computer, etc.), a plurality of realtime access requests (e.g., current access request 350). Each access request may comprise a plurality of data elements (e.g.. data elements 350B) and a resource identifier (e.g., resource identifier 350A) for the respective resource. The desired resource can include access to a secure location or building, a good, a service, etc.
[0113] In step 516, the current detector value for the first seed field and resource identifier for the respective resource are extracted for each real-time access request. The current detector value may be extracted by a subsystem (e.g., detector extractor 302) or other suitable means and transmitted to another subsystem for processing (e.g., segment module 304).
[0114] In step 518, the current detector value may be used to generate a current thread of category labels for the access request from the set of access requests. A subsystem (e.g., segment module 304) or other suitable means may generate the current thread of category labels for the current detector value by determining the count of unique instances the current detector value has been used in connection with another detector value.
[0115] In step 520, the current thread of category labels is matched to a first cluster thread of the set of cluster threads. The subsystem (e.g., comparison module 310B) or other suitable means may use the current thread of category labels as an identifier to identity7a matching cluster thread (e.g., cluster thread 420B-430B) from the table (e.g., historical access result index 400B) stored in the database (e.g.. database 306).
[0116] In step 522, an access request outcome is determined for the real-time access request using the one or more access metrics of the first cluster thread. The access request3079088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 outcome may be determined from the access metrics corresponding to the matching cluster thread identified in the table.
[0117] In step 524. access to the resource is provided based on the access request outcome. If the access request outcome indicates a valid access request, access to the resource identified by the resource identifier associated with the access request may be granted. For example, this may be access to a physical location (e.g., secure building) or access to an electronic account (e.g., debit or credit card account). Access may also be in the form of an authorization response to the resource computer (e.g., an issuer computer) that can authorize access to an electronic account.VI. COMBINATION OF GEOGRAPHICAL DETECTORS
[0118] As mentioned above, several benefits of using higher-dimensional input include increased accuracy when determining the validity of access request. Building upon the methods and techniques discussed above, cross combination of geographical detectors may determine the validity of an access request by iteratively analyzing the relationship between data elements within a vector associated with the access request. Specifically, beginning with the first data element of a vector, each element of a vector may be compared against itself and each previous element in the vector. Each pairwise comparison is assigned a category label and the collection of pairwise comparison labels for a given geographical detector is used to generate a segment thread. The set of segment threads generated for the set of detectors is used to generate a cluster thread. Because the geographical detector relationships are assigned a category label and each category label may be expressed by an abbreviation, the relationship between all the geographical detectors associated with an access request can therefore be represented efficiently as a single cluster thread.A. Generating segment thread and cluster threads
[0119] A sequence of geographical detector category labels representing the intrarelationship of geographical detectors may be generated to form a segment thread. The set of segment threads generated for a single access requests can be used to generate a cluster thread, which represents the high-dimensional input as a single string, thereby reducing the3179088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 computational burden of processing such information. One technique for generating a cluster thread representing the intra-relationship of geographical detectors is described below.
[0120] FIG. 6A illustrates an example geographical detector vector generated for a particular access request, in accordance to some embodiments. The system may receive an access request and the set of data corresponding to fields associated with the access request. The access request may stem from, as an example, a transaction such as an online credit card transaction. The access request and its corresponding data may be received from a plurality of devices (e.g.. user device, access device, request computer, etc.). At least three geographical detectors are selected from the set of fields associated with the access request based upon predetermined criteria. The criteria for selecting a set of geographical detectors may depend on the observed accuracy performance of such fields selected to be geographical detectors. The selected geographical detectors 606A-616A may be stored in any suitable data structure such as a vector 1 602A and vector 2604A. The geographical detector can be referenced by its position in the vector (e.g., 1, 2, 3, etc.) or by an abbreviation assigned to that vector (e.g., “B” for billing country , “S'’ for shipping country , “I” for IP country , etc.).
[0121] Additionally, geographical detectors may be generated from geographical information associated with the access request. For example, if an IP address is associated with the access request stemming from an online credit card transaction, then the longitude and latitude coordinates associated with the access request may be determined. A geographical detector may then be generated to correspond to “IP Address’" and the determined coordinates may be the value assigned to the generated geographical detector.
[0122] Beginning with the first geographical detector 606A, the geographical detector value is compared against itself and the result is recorded. The results may be recorded in a vector. Since there are no preceding geographical detectors, the result of 606A may be stored and the system proceeds to the next geographical detector 608A. Geographical detector 608A is compared against itself and preceding geographical detector 606A and each result is recorded. The results of the pairwise comparisons of 606A are stored together at the same index position. For instance, the index storing the results of 608 A may contain 2 category' labels corresponding to the comparison of 608A to itself and the comparison of 608A with 606 A. Next, geographical detector 610A is compared against itself, then against preceding geographical detector 608A, and finally against preceding geographical detector 606A. The3279088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 index storing the results of 610A’s comparisons therefore may contain 3 results. The process is repeated for each geographic detector in the set of geographic detectors. Therefore, for a data element located at a position n in the vector, the data element may be compared against each data element located at positions less than or equal to n, including itself. The comparison process proceeds incrementally through the vector until the last data element at the final position in the vector is compared against all preceding data elements and itself, thereby generating a complete set of pairwise comparison results. Therefore, the number of results stored in a given index position n can equal to n.
[0123] Each pairwise comparison is assigned a category label indicating the result of the pairwise comparison. The category' label may be predetermined and correspond to an abbreviation for a comparison result. For instance, the category labels may be “S", “D", “E”, and “N", which are abbreviations for ‘"same’; “different", “empty", and “not applicable", respectively.
[0124] FIG. 6B illustrates an example cluster thread 602B generated for an access request. Each segment thread 606B-616B corresponds to a unique geographical detector 606A-616A. The length of each segment thread 606B-616B may differ from each other because each geographical detector 606A-616A is comparing against one additional geographical detector than the previous geographical detector. For instance, the geographical detector 606A is the first geographical detector and only performs one pairwise comparison (against itself).Therefore, the segment thread corresponding to geographical detector 606A comprises one category label as depicted in segment thread 606B. Geographical detector 608A performs two comparisons: one comparison against itself and one comparison against geographical detector 606A. Therefore, the resulting segment thread 608B comprises two category labels.
[0125] Each segment thread of the set of segment threads in cluster thread 602B may be separated by a dash or other suitable marker. For example, a dashmay be appended to the end of the segment thread generated for geographical detector 606A and the segment thread for geographical detector 608A may be appended after the dash. In another example, the segment threads may be concatenated end-to-end and not separated by any dash or other identifier.3379088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 B. Generating and using a historical access result index
[0126] FIG. 7 is an example historical access result index 700 generated from cluster threads. The historical access result index 700 may be generated using the methods and embodiments disclosed herein. Each row in cluster thread column 702 may correspond to a cluster thread generated for a historical access request of a set of historical access request. Each column may contain a metric or statistical information characterizing the validity of the respective cluster thread. Metrics may include those such as percentage of access requests corresponding to that cluster thread that are reported as a fraudulent, false positive, chargeback, or authorized request. The metrics corresponding to each cluster thread may be updated when an outcome label assigned to an access request corresponding to the cluster thread is received, as described herein.
[0127] The resulting cluster thread (e g., cluster thread 602B) generated for a current access request may be used as an identifier to identify a matching cluster thread in the cluster thread column 702 of the historical access result index 700. Once a matching thread in the cluster column 702 is found, the corresponding metrics from columns 704-716 may be used to determine whether or not to grant access to the resource identified by the current access request. Access request may be determined by assessing if the metrics and / or values associated with the identified cluster thread exceeds a predetermined risk tolerance threshold and / or violate one or more rules. For example, if the cluster thread is associated with a chargeback percentage exceeding a particular threshold, the access request determination module may deem the current access request as unauthorized. As another example, a rule may stipulate an access request be denied if its corresponding cluster thread matches a cluster thread in cluster thread column 702 with a history of invalid requests exceeding a threshold percentage.C. Example
[0128] A cluster may be generated for every access request. For a given access request, at least three geographical detectors are selected from the set of fields associated with the access request. For example, an access request stemming from an online credit card transaction may contain several fields (e g., shipping country, billing country, email, account, etc.). From the set of fields, three or more geographical fields may be selected and determined to be the3479088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 geographical detectors. Geographical detectors may be, as an example, the billing country', bank identification number (BIN) country, shipping country, currency country, actual IP country (country the transaction is originating from), and proxy IP country (country- presented during the transaction), e.g., as depicted by geographical detectors 606A-616A. The geographical detectors may be stored in a vector (e.g., vector 602A-602B) yvith each detector belonging to a unique index position within the vector. References made to a particular geographical detector may be done so by referencing its assigned index number.
[0129] The category labels may refer to the comparison result. For this example, the category labels may correspond to a comparison result such as: ‘‘same”, “different”, “empty”, or “not applicable” and abbreviated to “S”, “D”, “E”, and “N”, respectively.
[0130] Referring to FIG. 6A, vector 1 602A and vector 2602B are exemplar}- geographical detector vectors generated and corresponding to 2 unique access requests. Starting with vector 1 602 A and beginning with geographical detector 606A, the data element associated with geographical detector 606A, “United States”, is compared with itself. Since no other data elements precede geographical detector 606A, only one category- label result may be recorded in the thread vector. Comparisons of a valid data element against itself will be labeled “S” for same. Therefore, the resulting segment thread for geographical detector 606A is “S”.
[0131] Vector 1 602A geographical detector 608A is data element “Portugal”. “Portugal” is a valid data element and results in category' label “S” yvhen compared to itself and “D” yvhen compared to preceding geographical detector 608A data element “United States”. Therefore, the resulting segment thread for geographical detector 608A is “DS”.
[0132] Vector 1 602A geographical detector 610A is data element “Unknown”.“Unknown”, in this example”, is not a valid data element and indicates a missing data element. Therefore, yvhen compared to itself, the resulting category label is “E” to denote “empty”. The resulting category label from comparing geographical detector 610A to geographical detector 608A is “N” for not applicable. The same step is repeated for geographical detector 606A. The resulting segment thread for geographical detector 610A is “NNE”.3579088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01
[0133] Vector 1 602A geographical detector 612A is data element "United Kingdom”, which is a valid data element in this example. The resulting segment thread for geographical detector 612A, repeating the steps as described above, is “DDNS”.
[0134] Vector 1 602A geographical detector 614A is data element “Germany”, which is a valid data element in this example. The resulting segment thread for geographical detector 614A, repeating the steps as described above, is “DDNDS”.
[0135] Vector 1 602A geographical detector 616A is data element “Unknown”, which is an invalid data element in this example. The resulting segment thread for geographical detector 616A, repeating the steps as described above, is “NNNNNE”.
[0136] The set of segment threads generated for vector 1 602 A may then combined to generate a cluster thread (e.g, 602B). A dash (“-“) may separate each segment thread.
[0137] Repeating the steps described herein for vector 2604A will result in the same segment threads, thereby generating the same cluster thread (e.g., 602B).
[0138] The cluster thread 602B of FIG. 6B, is an example cluster thread corresponding to vector 1 602A and vector 2604 A. Segment threads 606B-616B correspond to the segment threads generated for geographical detectors 606A-61 A. In this example, the segment threads 606B-616B have been concatenated together with each segment thread delimited by a dash. The first segment thread 606B may not be preceded by a delimiter.
[0139] Referring to FIG. 7, the resulting geographical detector thread 602B may be used to identify a matching clsuter thread stored in the cluster thread column 702. The corresponding metrics stored in columns 704-716 may be used to determine the validity of the access request associated with cluster thread 602B.D. Method
[0140] FIG. 8 depicts a flow chart of a method 800 for securing access to resources by determining the validity of an access request in accordance with an embodiment of the disclosure. The method may be performed by a detection mechanism (e.g., detection mechanism 300) or other suitable means. The detection mechanism can validate incoming access requests in real-time.3679088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01
[0141] In step 802, a set of access requests (e.g., access requests 301) may be received over a network from a plurality of devices. Each of the access requests may be associated with a resource identifier (e.g., resource identifier 301A) for a respective resource and a plurality of geographical detector values corresponding to at least three fields (e.g., resource country, IP country', currency country', etc.). The plurality' of geographical detector values may be identified from, for example, the data elements 301 B associated with the access request. The resource identifier and plurality of geographical detector values may be extracted by a subsystem (e.g., detector extractor 302) to be stored and transmitted as a vector. The set of access requests may include at least 500,000 access requests.
[0142] The at least three fields may be selected from: billing country', bank identification number (BIN) country, resource country (country where the resource is located, such as a physical building), shipping country, currency country, actual IP country (country the transaction is originating from), and proxy IP country (country presented during the transaction).
[0143] In step 804, the access result values for the set of access requests (e.g., historical access request 301) may be stored in a database (e g., database 306) accessible by the detection mechanism. The access result values may be transmitted as a Boolean, text string, or other suitable means (e.g., outcome labels 308) from a device or other suitable means (e.g., resource computer 320). The access result value may provide a measure of approval in granting valid access to the respective resource. In other words, the access result values may indicate past approved or rejected access requests.
[0144] In step 806. a pairwise comparison between each pair of the plurality of geographical detectors may be performed. The comparison can be performed by a subsystem (e.g., segment module 304). Each pairwise comparison may provide a category label indicating whether a match exists between the pair of geographical detectors. The category labels may be predetermined values that indicate the result of the pairwise comparisons.
[0145] The category label may correspond to a pairwise comparison result. Additionally, the pairwise comparison result may correspond to one of: ’‘same”, “different”, “empty”, and “not available”.3779088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01
[0146] In step 808, a thread of category7labels is generated for the geographical detectors, thereby generating a set of segment threads (e.g., 606B-616B). The thread of category labels may indicate one or more pairwise comparison results (e.g.. “same”, “different”, etc.). The category labels may comprise one or more characters (e.g., “S”, “D”, etc.) that is an abbreviation of the pairwise comparison result. The category7labels associated with an access request may be stored in a vector. The set of segment threads generated from the thread of category labels may be stored in a vector and each segment thread of the set of segment thread may be stored at an index position that indicates the geographical detector the segment thread corresponds to. The vector may be transmitted to a database (e.g., database 306) where it may be sorted into a searchable table (e.g., historical access result index 700). In some embodiments a subsystem (e.g., segment module 304) may handle this step.
[0147] The vector storing the category labels associated with an access request may also contain the resource identifier associated with the access request.
[0148] The segment threads of a set of segment threads can be concatenated together and each segment thread, except for the first segment thread, can be preceded by a character value. For example, a set of segment threads may contain 5 segment threads. Each of the 5 segment threads may be concatenated together by appending a segment thread to the end of a preceding segment thread and separating each segment thread with a character, such as a dash (“-“)•
[0149] In step 810, the detection mechanism 300 may determine one or more access metrics from the access result values (e.g., outcome labels 308) received for each cluster thread in a set of cluster threads. The access result value may be received from a device (e.g., resource computer 320) or other suitable means. The access results may be stored in a table (e.g., historical access result index 700) and each cluster thread may correspond to a row in the table. Each access result value may correspond to a cluster thread and may be stored in a column (e.g., cluster thread column 702) associated with the cluster thread. The access metncs may be determined by calculating metrics, such as percentage of invalid access results corresponding to the access request, based on the access result values. The database may update the table with the determined access metrics. The access metric may be periodically updated based on a new access value received.3879088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01
[0150] The cluster threads may be transmitted to a database where it may be sorted into a searchable table. The historical access result index may comprise a plurality of rows, each row corresponding to a cluster thread of the set of cluster threads and a plurality of columns, each column representing an access metric associated with the corresponding cluster thread.
[0151] In step 812, the detection mechanism 300 may receive, over a network from a plurality of devices (e.g., user device, an access device, request computer, etc.), a plurality7of real-time access requests (e.g., current access request 350 of FIG. 3B). Each access request may comprise a plurality of data elements (e.g., data elements 350B) and a resource identifier (e.g., resource identifier 350A) for the respective resource. The desired resource can include access to a secure location or building, a good, a sendee, etc.
[0152] In step 814, a plurality7of geographical detector values and the resource identifier may be extracted for each real-time access request. The plurality7of geographical detector values may be extracted from the data elements associated with the real-time access request.
[0153] In step 816, a current thread of category labels may be generated using the plurality of geographical detector values. A subsystem (e.g., segment module 304) may generate the category labels by determining the result of a pairwise comparison for a given geographical detector, thereby generating a segment thread (e.g., 606B-616B) for the given detector. The set of segment threads is used to generate a cluster thread (e.g., cluster thread 602B) for the given access request.
[0154] In step 818, the current thread is matched to a first cluster thread of the set of cluster threads. The current thread may be received by a subsystem (e.g., historical access result index module 310) or other suitable means.. The subsystem can identify, using the current thread as an identifier, a matching cluster thread of the set of cluster threads within table.
[0155] In step 820, an access request outcome is determined for the real-time access request using the one or more access metrics of the first segment thread. The access request outcome may be determined by identity ing, from a matching cluster thread the table’s column (e.g., cluster thread column 702), the corresponding metrics (e.g., 704-716).
[0156] In step 822. access to the resource is provided based on the access request outcome. If the access request outcome indicates a valid access request, access to the resource identified by the resource identifier associated with the access request may be granted. For3979088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 example, this may be access to a physical location (e.g., secure building) or access to an electronic account (e.g., debit or credit card account). Access may also be in the form of an authorization response to the resource computer (e.g., an issuer computer) that can authorize access to an electronic account.VII. COMPUTER SYSTEM
[0157] Any of the computer systems mentioned herein may utilize any suitable number of subsystems. Examples of such subsystems are shown in FIG. 9 in computer system 900. In some embodiments, a computer system includes a single computer apparatus, where the subsystems can be the components of the computer apparatus. In other embodiments, a computer system can include multiple computer apparatuses, each being a subsystem, with internal components. A computer system can include desktop and laptop computers, tablets, mobile phones and other mobile devices.
[0158] The subsystems shown in FIG. 9 are interconnected via a system bus 930.Additional subsystems such as a printer 908, keyboard 918, storage device(s) 920, monitor 914 (e.g., a display screen, such as an LED), which is coupled to display adapter 912. and others are shown. Peripherals and input / output (I / O) devices, which couple to I / O controller 902, can be connected to the computer system by any number of means known in the art such as input / output (I / O) port 916 (e.g., USB, FireWire®). For example, I / O port 916 or external interface 922 (e.g., Ethernet, Wi-Fi, etc.) can be used to connect computer system 900 to a wide area network such as the Internet, a mouse input device, or a scanner. The interconnection via system bus 930 allows the central processor 906 to communicate with each subsystem and to control the execution of a plurality of instructions from system memory 904 or the storage device(s) 920 (e.g., a fixed disk, such as a hard drive, or optical disk), as well as the exchange of information between subsystems. The system memory 904 and / or the storage device(s) 920 may embody a computer readable medium. Another subsystem is a data collection device 910, such as a camera, microphone, accelerometer, and the like. Any of the data mentioned herein can be output from one component to another component and can be output to the user.
[0159] A computer system can include a plurality of the same components or subsystems, e.g., connected together by external interface 922. by an internal interface, or via removable4079088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 storage devices that can be connected and removed from one component to another component. In some embodiments, computer systems, subsystem, or apparatuses can communicate over a network. In such instances, one computer can be considered a client and another computer a server, where each can be part of a same computer system. A client and a server can each include multiple systems, subsystems, or components. In various embodiments, methods may involve various numbers of clients and / or servers, including at least 10, 20, 50, 100, 200, 500. 1,000, or 10,000 devices. Methods can include various numbers of communication messages between devices, including at least 100, 200, 500, 1,000, 10,000, 50,000, 100,000, 500,00, or one million communication messages. Such communications can involve at least 1 MB, 10 MB, 100 MB, 1 GB, 10 GB, or 100 GB of data.
[0160] Aspects of embodiments can be implemented in the form of control logic using hardware circuitry (e g., an application specific integrated circuit or field programmable gate array) and / or using computer software stored in a memory' with a generally programmable processor in a modular or integrated manner, and thus a processor can include memory storing software instructions that configure hardware circuitry, as well as an FPGA with configuration instructions or an ASIC. As used herein, a processor can include a single-core processor, multi-core processor on a same integrated chip, or multiple processing units on a single circuit board or networked, as well as dedicated hardware. Based on the disclosure and teachings provided herein, a person of ordinary skill in the art will know and appreciate other ways and / or methods to implement embodiments of the present disclosure using hardware and a combination of hardware and software.
[0161] Any of the software components or functions described in this application may be implemented as software code to be executed by a processor using any suitable computer language such as. for example, Java, C, C++, C#, Objective-C, Swift, or scripting language such as Perl or Python using, for example, conventional or object-oriented techniques. The software code may be stored as a series of instructions or commands on a computer readable medium for storage and / or transmission. A suitable non-transitory computer readable medium can include random access memory (RAM), a read only memory (ROM), a magnetic medium such as a hard-drive or a floppy disk, or an optical medium such as a compact disk (CD) or DVD (digital versatile disk) or Blu-ray disk, flash memory, and the like. The4179088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 computer readable medium may be any combination of such devices. In addition, the order of operations may be re-arranged. A process can be terminated when its operations are completed but could have additional steps not included in a figure. A process may correspond to a method, a function, a procedure, a subroutine, a subprogram, etc. When a process corresponds to a function, its termination may correspond to a return of the function to the calling function or the main function.
[0162] Such programs may also be encoded and transmitted using carrier signals adapted for transmission via wired, optical, and / or wireless networks conforming to a variety of protocols, including the Internet. As such, a computer readable medium may be created using a data signal encoded with such programs. Computer readable media encoded with the program code may be packaged with a compatible device (e.g., as firmware) or provided separately from other devices (e.g., via Internet download). Any such computer readable medium may reside on or within a single computer product (e g., a hard drive, a CD, or an entire computer system), and may be present on or within different computer products within a system or network. A computer system may include a monitor, printer, or other suitable display for providing any of the results mentioned herein to a user.
[0163] Any of the methods described herein may be totally or partially performed with a computer system including one or more processors, which can be configured to perform the steps. Any operations performed with a processor may be performed in real-time. The term “real-time” may refer to computing operations or processes that are completed within a certain time constraint. As examples, a time constraint may be 30 seconds, 1 minute, 10 minutes, 30 minutes, 1 hour, 4 hours, 1 day, or 7 days. Thus, embodiments can be directed to computer systems configured to perform the steps of any of the methods described herein, potentially with different components performing a respective step or a respective group of steps. Although presented as numbered steps, steps of methods herein can be performed at a same time or at different times or in a different order. Additionally, portions of these steps may be used with portions of other steps from other methods. Also, all or portions of a step may be optional. Additionally, any of the steps of any of the methods can be performed with modules, units, circuits, or other means of a system for performing these steps.
[0164] The specific details of particular embodiments may be combined in any suitable manner without departing from the spirit and scope of embodiments of the disclosure.4279088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 However, other embodiments of the disclosure may be directed to specific embodiments relating to each individual aspect, or specific combinations of these individual aspects.
[0165] The above description of example embodiments of the present disclosure has been presented for the purposes of illustration and description. It is not intended to be exhaustive or to limit the disclosure to the precise form described, and many modifications and variations are possible in light of the teaching above.
[0166] A recitation of "a", "an" or "the" is intended to mean "one or more" unless specifically indicated to the contrary. The use of "or" is intended to mean an “inclusive or / ’ and not an “exclusive or” unless specifically indicated to the contrary. Reference to a “first” component does not necessarily require that a second component be provided. Moreover, reference to a “first” or a “second” component does not limit the referenced component to a particular location unless expressly stated. The term “based on” is intended to mean “based at least in part on.”
[0167] The claims may be drafted to exclude any element which may be optional. As such, this statement is intended to serve as antecedent basis for use of such exclusive terminology as “solely”, “only”, and the like in connection with the recitation of claim elements, or the use of a “negative” limitation.
[0168] All patents, patent applications, publications, and descriptions mentioned herein are incorporated by reference in their entirety for all purposes. None is admitted as prior art. Where a conflict exists between the instant application and a reference provided herein, the instant application shall dominate.4379088584V.5
Claims
PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 WHAT IS CLAIMED IS:
1. A method for securing access to resources, the method comprising performing, by a computer system:receiving, over a network from a plurality7of devices, a set of access requests, each including a resource identifier for a respective resource and a plurality7of detector values corresponding to at least three fields, the set of access requests including at least 500.000 access requests;storing access result values for the set of access requests, wherein the access result values provide a measure of approval in granting valid access to the respective resource; selecting a first field of the at least three fields to be a first seed field;for each particular detector value of the first seed field:determining a category label for other fields of the at least three fields, wherein the category label indicates an amount of different detector values in access requests that include the particular detector value of the first seed field;generating a segment thread of category labels for the particular detector value, thereby generating a cluster thread ;for each cluster thread of a set of cluster threads, determining one or more access metrics from the access result values;receiving, over a network from a plurality of devices, a plurality of real-time access requests;for each real-time access request:extracting (1) a current detector value for the first seed field and (2) the resource identifier for the respective resource;generating, using the current detector value, a current cluster thread of category labels from the set of access requests;matching the current cluster thread to a first cluster thread of the set of cluster threads;determining an access request outcome for the real-time access request using the one or more access metrics of the first cluster thread; andproviding access to the resource based on the access request outcome.4479088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 2. The method of claim 1, wherein the plurality of detector values corresponds to at least seven fields.
3. The method of claim 1, further comprising storing the one or more access metrics from the access result values for each cluster thread in a historical access result index, the historical access result index comprising:a plurality of rows, each row corresponding to a cluster thread of the set of cluster threads; anda plurality of columns, each column representing an access metric associated with the corresponding cluster thread.
4. The method of claim 1 , further comprising:selecting a second field of the at least three fields to be a second seed field;for each particular detector value of the second seed field:determining a category label for other fields of the at least three fields, wherein the category label indicates an amount of different detector values in access requests that include the particular detector value of the second seed field;generating a second segment thread of category labels for the particular detector value;generating the cluster thread by concatenating the segment thread and second segment thread.
5. The method of claim 1, wherein the access request outcome indicates denial of access to the resource based on the one or more access metrics exceeding a threshold value.
6. The method of claim 1 , wherein the category label corresponds to a range of values, the range being specified by an upper and lower boundary values.
7. The method of claim 1, wherein the category labels associated with an access request are stored in a vector, wherein the vector contains the resource identifier associated with the access request.
8. The method of claim 1, wherein the one or more access metrics are updated periodically based on new access values received.4579088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 9. The method of claim 1 , wherein the at least three fields include at least one selected from email, user device, IP address, and phone number.
10. The method of claim 1, wherein the first seed field is selected based on a ranking of importance values of fields in predicting the validity7of an access result.
11. A method for securing access to resources, the method comprising performing, by a computer system:receiving, over a network from a plurality7of devices, a set of access requests, each including a resource identifier for a respective resource and a plurality of geographical detector values corresponding to at least three fields, the set of access requests including at least 500,000 access requests;storing access result values for the set of access requests, wherein the access result values provide a measure of approval in granting valid access to the respective resource; for each access request of the set of access requests:performing a pairwise comparison between each pair of the plurality of geographical detectors, wherein each pairwise comparison provides a category label indicating whether a match exists between the pair of geographical detectors, thereby- generating a set of segment threads of category labels; andgenerating a cluster thread from the set of segment threads of category labels. ; for each cluster thread of a set of cluster threads, determining one or more access metrics from the access result values;receiving, over a network from a plurality7of devices, a plurality of real-time access requests;for each real-time access request:extracting (1) the plurality of geographical detector values and (2) the resource identifier for the respective resource;generating, using the plurality of geographical detector values, a current cluster thread of category labels;matching the current thread to a first cluster thread of the set of cluster threads; determining an access request outcome for the real-time access request using the one or more access metrics of the first cluster thread: andproviding access to the resource based on the access request outcome.4679088584V.5PATENT Attorney Docket No. 079900-1475257-9718W001Client Ref. No. 9718WO01 12. The method of claim 11, further comprising storing the one or more access metrics from the access result values for each cluster thread in a historical access result index, the historical access result index comprising:a plurality of rows, each row corresponding to a cluster thread of the set of cluster threads; anda plurality of columns, each column representing an access metric associated with the corresponding cluster thread.
13. The method of claim 11 wherein a category' label corresponds to a pairwise comparison result.
14. The method of claim 13, wherein the pairwise comparison result indicates one of the geographical detectors are the same, different, empty, or not available.
15. The method of claim 11, wherein the category labels associated with an access request are stored in a vector, wherein the vector contains the resource identifier associated with the access request.
16. The method of claim 11, wherein the access metric is updated periodically based on new access values received.
17. The method of claim 11, wherein at least one of the at least three fields is selected from resource country, actual IP country, and proxy IP country.
18. The method of claim 11, wherein generating the cluster thread includes concatenating the segment threads of the set of segment threads and w herein each segment thread, except for the segment thread, is preceded by a designated character value.
19. A computer system for securing access to resources, the computer system comprising:one or more processors;a non-transitory computer readable medium storing a plurality of instructions that, w hen executed, control the one or more processors to perform the method of any one of claim 1-18.4779088584V.5
