System and method for device identification and uniqueness

Abstract

Systems and methods for determining uniqueness of device identifiers are provided are provided. The uniqueness of a device identifier may be indicated by a device quality score or grade that is calculated based on a plurality of parameters associated with a device identifier as well as evaluation rules derived based on historical data. The plurality of parameters may be associated with a network event or transaction associated with the device identifier. The evaluation rules may be derived using machine learning techniques. Based on uniqueness of a device identifier, a suitable action or measure may be taken.

Description

BACKGROUND OF THE INVENTION

With rapid advancement of computer technologies and e-Commerce, people are increasingly reliant on a variety of internet-connected devices for everything from banking to booking travel to shopping. As a service provider, it has become increasingly important to distinguish among the different devices in order to provide detect and prevent online fraud and/or to provide customized content or services.

SUMMARY OF THE INVENTION

System and methods for determining uniqueness of device identifiers are provided. According to an aspect of the invention, a computer-implemented method for determining uniqueness of a device identifier is provided. The method comprises obtaining a plurality of evaluation rules based at least in part on historical data and determining the uniqueness of the device identifier based at least in part on the evaluation rules and a plurality of parameters associated with the device identifier. Obtaining the one or more evaluation rules may include analyzing the historical data using a machine learning technique. Determining the uniqueness of the device identifier may include determining a device quality score associated with the device identifier. Determining the uniqueness of the device identifier may include selecting a subset of the one or more evaluation rules based at least in part on the plurality of parameters and applying the subset of evaluation rules to at least some of the plurality of parameters to obtain the device quality score. The method may further comprise determining a suitable action based at least in part on the uniqueness of the device identifier. Determining the suitable action may include selecting a first action if the device identifier is more likely to be unique and selecting a second action if the device identifier is less likely to be unique.

INCORPORATION BY REFERENCE

All publications, patents, and patent applications mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent, or patent application was specifically and individually indicated to be incorporated by reference.

BRIEF DESCRIPTION OF THE DRAWINGS

The novel features of the invention are set forth with particularity in the appended claims. A better understanding of the features and advantages of the present invention will be obtained by reference to the following detailed description that sets forth illustrative embodiments, in which the principles of the invention are utilized, and the accompanying drawings of which:

FIG. 1 illustrates an example environment for implementing the present invention, in accordance with an embodiment.

FIG. 2a illustrates example components of a device identification system, in accordance with an embodiment.

FIG. 2b illustrates example components of a device identification system, in accordance with another embodiment.

FIG. 3 illustrates example components of a computer device for implementing aspects of the present invention, in accordance with an embodiment.

FIG. 4 illustrates an example process for implementing the present invention, in accordance with an embodiment.

FIG. 5 illustrates an example process for determining the uniqueness of a device identifier, in accordance with an embodiment.

FIG. 6 illustrates an example process for determining the uniqueness of a device identifier, in accordance with an embodiment.

FIG. 7 illustrates an example process for calculating a device quality score, in accordance with an embodiment.

FIG. 8 illustrates an example process for determining the uniqueness of a device identifier, in accordance with an embodiment.

DETAILED DESCRIPTION OF THE INVENTION

According to aspects of the present invention, a device identification system and methods may be provided for determining the uniqueness of device identifiers. Network devices, such as desktops, laptops, tablet computing devices, smart phones, smart TVs, and the like, may be identified using device identifiers. Such device identifiers may be provided or generated by the manufacturers, distributors, developers, or any suitable entity. Examples of device identifier may include Android identifier (ID), iPhone's Unique Identifier (UDID), iPhone's IdentifierForAdvertising (IFA or IDFA), cookie ID, login ID, Internet Protocol (IP) address, media access control (MAC) address, a hash of any of the above, a combination of any of the above, or the like. In some cases, the device identifier may be derived based on one or more hardware and/or software parameters of a device identified by the device identifier. For example, a device identifier may be derived from the IP address, operating system version, and locale setting of the device. In some embodiments, a device identifier may be used to identify the source or origin or a transaction, request, or network event. For example, a device identifier may include a user identifier, an account identifier, and the like. Ideally, a device identifier uniquely identifies a device. In other words, different devices are mapped to different device identifiers, for example, based on unique software/hardware characteristics associated with the devices. However, in some cases, different devices may have the same device identifiers. In some cases, such as in an online fraud, such sharing of device identifiers may be intentional. In some other cases, such sharing of device identifiers may be unintentional.

In some embodiments, a device identifier may be used to distinguish among the entities (e.g., users) associated with the devices identified by the device identifier. For example, a content or service provider may use such device identifiers to distinguish among different users so as to provide customized advertisement or service items tailored to the preferences of the users. However, such targeted user-specific action (e.g., target advertisement delivery) is only effective when there is a high probability that the device identifier is indeed unique, that is, it is very likely that the device identifier is not shared by multiple devices. In some instances, different devices may have the same device identifiers. For example, the Android ID for two Android devices may be the same. For another example, two devices may have the same IP address. As yet another example, devices having similar or different parameters may result in having the same value for their device identifiers. In such cases when the device identifier is not unique, targeted action may not be feasible or desirable. Rather, a different action or approach may be required.

As an example, consider two users Carola and Marley. Carola operates a device 1 to access a shopping website and Marley operates a device 2 to access the same shopping site. Carola is interested in jewelry and art and Marley is interested in cars and technology. The service provider operating the shopping website may be able to deliver different targeted advertisement to Carola and Marley based on their different preferences if the service provider can distinguish the device 1 from device 2. In other words, given a device identifier, if the service provider can be confident that the device identifier is unique, that is, it is unique to a specific device, then the service provider's may be able to provide effective targeted content. On the other hand, if for a given the device identifier, the service provider is not confident that the device identifier is unique (i.e., associated with only with Carola or only with Marley), then it may be undesirable to delivery customized content because it may offend or otherwise alienate the unintended audience of such customized content. Rather, providing more generic content may prove to be more effective in this case.

As another example, a fraud detection engine may be configured to detect fraudulent activities associated of network devices. For a given device identifier associated with fraudulent transactions, if it is likely that the device identifier is unique, it may be desirable to take specific fraud prevention actions with respect to activities associated with the device identifier. For example, the device identifier may be added to a blacklist so that activities associated with the device identifier are blocked. This may be a desirable solution that prevents future fraud without affecting non-rogue devices since it is unlikely that the device identifier is shared by other devices. Conversely, if the device identifier is likely to be non-unique, then put the device identifier into a blacklist may unduly impact legitimate activities of non-rogue devices that happen to share the same device identifier. Rather, a more moderate measure may be taken such as monitoring instead of blocking traffic coming from devices identified by the device identifier.

According to aspects of the present invention, a device identification system and methods may be provided for determining the uniqueness of device identifiers. The device identification system may include an analysis engine, an evaluation engine, and optionally, an action engine (such as illustrated in FIG. 2a below). The analysis engine may be configured to generate, based on historical data, rules that may be used to determine the uniqueness of device identifiers. Specifically, the rules may be used to derive a quality score associated with a given device identifier. Quality score may be a numerical value. For instances, the quality score may be any integer between 0 and 100 (inclusive) where the higher the quality score the more likely that the device identifier is unique. The rules may be stored in a data store which may include a database or a data file (e.g., a JSON file) that is accessible to the evaluation engine.

The evaluation engine may be configured to receive a plurality of parameters associated with a device identifier, select the applicable rules (e.g., based on at least some of the parameters such as the device's operating system or hardware type) and apply the rules to the plurality of parameters to derive a quality score and/or a quality grade. In some cases, the quality grade may be derived based on the quality score. In some embodiments, the plurality of parameters may include any hardware and/or software parameters associated with a device identified by the device identified.

In some embodiments, the plurality of parameters may include deterministic identifiers such as customer ID, login ID, account ID, cookie ID, UDID, Android ID, IFA or IFDA, Identifier For Vendor (IFV), International Mobile Equipment Identity (IMEI), MAC address, IP address, and the like.

In some embodiments, the plurality of parameters may include client and/or server location and/or geographical information; client device trustability score or similar indicator; user lifetime value (LTV) or similar indicator; client device Return on Investment Index (ROlndex) or similar indicator; device metadata such as manufacturer (e.g., Apple, Samsung, Microsoft, Dell, etc), name (e.g., iPhone), model, version, and the like; device operating system (e.g., iOS, Android, Windows Phone, BlackBerry, Mac OS, OS X, Microsoft Windows, Unix, Linux, BSD, etc.); browser metadata such as maker (e.g., Google, Microsoft, Mozilla), name (e.g., Chrome, Internet Explorer, Firefox, Opera, Safari), version, and the like; other software and/or hardware characteristics (e.g., Adobe Flash version); event type such as impression, click or selection, download, page load, and the like; event type detail such as campaign type, transaction type, and the like; customer type and industry; and other parameters.

In typical embodiments, such parameters are obtained without the awareness of the user operating the device. In some cases, some or all of the parameters may be obtained with the user's awareness. In some embodiments, some or all of the parameters may be included in one or more requests or messages provided by the device or by another entity such as a Domain Name System (DNS) server, an Active Directory (AD) server, and the like. Example parameters may include user agent (UA), IP address, user identity, user credentials, network protocol, service endpoint, Hypertext Transfer Protocol (HTTP) method and/or status code, operating system, locale or language code, processor architecture, device type (e.g., desktop, mobile phone, etc.) and the like. In some embodiments, some or all of the parameters discussed herein may be used to derive the device identifier and to derive the quality score or grade. For example, the parameters may be concatenated, combined, appended, hashed, encrypted, and otherwise processed to derive the device identifier and/or quality score or grade. For example, in an embodiment, the device identifier may include a 40-character SHA-1 hash of some of the parameters.

In some embodiments, a set of applicable rules may be selected based at least in part on some of the parameters associated with the device identifier. For example, different sets of rules or the same set of rules may be provided for different types of devices and/or operating systems. For example, a first set of rules may be selected for an Android device whereas a second set of rules may be selected for an iOS device.

In an embodiment, the rules include, for each of at least some of the plurality of parameters, a corresponding value-weight map. In some cases, the rules may include usage rules associated with the value-weight maps that specify how the maps should be applied to the parameter. In other embodiments, such usage rules may be optional. Each of the value-weight maps may include one or more parameter values or value ranges along with corresponding weights. The parameter values may include any of the above-discussed parameters or a derivation thereof. The table below provides an example value-weight map for the user agent parameter:

Parameter Value Weight
8               0.5
10              0.38
12              0.01
. . .           . . .

For the above example, a usage rule associated with the value-weight map may specify that the parameter value to be used to look up the value-weight map is the length of the user agent identifier character string. For example, given a user agent parameter of “UA_2.8.1”, the corresponding parameter value, according to the “UA-length” rule, is 8, the length of the character string “UA_2.8.1” and the corresponding weight is 0.5.

Thus, for at least some of the plurality of parameters associated with the device identifier, corresponding weights may be obtained by applying the rules (e.g., by looking up the corresponding parameter-specific value-weight map as specified by usage rule).

While the value-weight maps illustrated here each corresponds to a specific parameter, in some embodiments, a value-weight map may correspond to more than one parameter. For example, the value used to look up the value-weight map may be derived based on the values of one or more parameters.

In some embodiments, the weights associated with the parameters may be weighted, for example, based on the perceived importance of the parameters. The perceived importance of the parameters may be determined based on statistical analysis of the historical data. For example, in an embodiment, the weight associated with the IP address parameter may be given a larger weight than the weight associated with the user agent parameter.

A quality score indicative of the uniqueness of a device identifier may be derived based on the parameter weights, which may be weighted as discussed above. For example, the device score may be calculated as a linear combination of the weighted weight values. The quality score may be categorized into quality grades. For example, a quality score between 80 and 100 may be categorized as quality grade A, a quality score between 60 and 80 may be categorized as quality grade B, and so on. Thus, device identifiers may be segmented according to their quality grades. In general, a device identifier is considered to be “high quality” if it has a high quality score or grade and “low quality” if it has a low quality score or grade.

In some embodiments, the value-weight maps, usage rules, formula and/or algorithm for calculating the quality scores and the like are collectively referred to as the evaluation rules (or rules). Some or all of such evaluation rules may be derived based on the historical data associated with past user activities and usage of computing resources. In particular, the historical data may be analyzed using statistical analysis and machine learning techniques such as logistical regression. Other suitable data mining techniques may also be used. Such data analysis may be performed by the analysis engine with or without human intervention.

As discussed above, the calculation of the quality scores or grades does not involve analyzing vast amount of historical data. Rather, the calculation is performed based on the rules derived from the historical data. The size of the rules may be significantly smaller than the size of the historical data which the rules are based on. For example, the rules may fit in one or more JSON files whereas the historical data may be stored in large data storage systems. Given the pre-calculated or derived rules, the time and complexity of the quality score/grade calculation (and hence determination of uniqueness of device identifiers) is significantly reduced. In some cases, uniqueness determination may be performed efficiently for a large amount of transactions in a short period of time. As the historical data evolve over time, the rules derived from the historical data may be updated to reflect any changes (e.g., on a periodic basis). By using such updated rules, the quality score/grade calculation also reflects the changes in historical data.

In some embodiments, the device identification system discussed herein may or may not include an action engine. The action engine may be configured to take different actions based on the uniqueness of device identifiers (such as indicated by the quality grades or quality scores). For example, the action engine may be configured to provide more targeted content to devices associated with a high quality device identifier and less targeted content to devices associated with a lower quality device identifier. It shall be understood that different aspects of the invention can be appreciated individually, collectively, or in combination with each other.

FIG. 1 illustrates an example environment 100 for implementing the present invention, in accordance with an embodiment. As illustrated, one or more user devices 102 connect via a network 104 to a device identification system 106 configured to provide device identification functionalities described herein. In various embodiments, the user devices 102 may include any devices capable of communicating with the network 104, such as personal computers, workstations, laptops, smartphones, mobile phones, tablet computing devices, smart TVs, game consoles, internet-connected setup boxes, kitchen appliances and the like. In some embodiments, the user devices 102 may include applications such as web browsers and/or applications (e.g., mobile apps) that are capable of communicating with the device identification system 106 and/or a system that uses the device identification system 106.

In some embodiments, the device identification system 106 may include or be included in one or more computing systems. For example, the device identification system 106 may be a part of a content provider. For example, the device identification system 106 may be a runtime component of a web server of the content server. As another example, the device identification system 106 may be a part of a fraud detection system or service used by an online service provider such as a bank, a merchant, a payment service provider, and the like. In some embodiments, the device identification system may be owned and/or operated by the same or different entity as the content provider.

In some embodiments, the device identification system 106 may be implemented by one or more physical and/or logical computing devices or computer systems that collectively provide the functionalities described herein. For example, aspects of the device identification system 106 may be implemented by a single server or by a plurality of servers (e.g., distributed Hadoop nodes). As another example, aspects of the device identification system 106 may be implemented by one or more processes running on one or more devices. In some embodiments, the device identification system 106 may provide an API such as a web service interface that may be used by users or other processes or services to utilize the functionalities of the device identification system discussed herein.

In some embodiments, the device identification system 106 may comprise one or more computing services provisioned from a “cloud computing” provider, for example, Amazon Elastic Compute Cloud (“Amazon EC2”), provided by Amazon.com, Inc. of Seattle, Wash.; Sun Cloud Compute Utility, provided by Sun Microsystems, Inc. of Santa Clara, Calif.; Windows Azure, provided by Microsoft Corporation of Redmond, Wash., and the like.

In some embodiments, the device identification system 106 may communicate with a data store 108 in order to perform the functionalities described herein. For example, the data store 108 may be used to store historical data, evaluation rules, and the like.

In some embodiments, the data store 108, or any other data stores discussed herein, may include one or more data files, databases (e.g., SQL database), data storage devices (e.g., tape, hard disk, solid-state drive), data storage servers, or the like. In various embodiments, such a data store 108 may be connected to the device identification system 106 locally or remotely via a network. In some embodiments, data store 108, or any other data stores discussed herein, may comprise one or more storage services provisioned from a “cloud storage” provider, for example, Amazon Simple Storage Service (“Amazon S3”), provided by Amazon.com, Inc. of Seattle, Wash., Google Cloud Storage, provided by Google, Inc. of Mountain View, Calif., and the like.

In various embodiments, the network 104 may include the Internet, a local area network (“LAN”), a wide area network (“WAN”), a cellular network, wireless network or any other public or private data and/or telecommunication network.

FIG. 2a illustrates example components of a device identification system 200A, in accordance with an embodiment. The device identification system 200A may be similar to the device identification system 106 discussed in FIG. 1. In various embodiments, the device identification system 200A may include one or more components that individually or collectively provide a set of functionalities. Each component may be implemented by one or more physical and/or logical computing devices, such as computers, data storage devices and the like. Some or all of the components may be co-located on the same device or distributed on different devices. The components may communicate with each other or with external entities such as other systems, devices or users. It will be appreciated by those of ordinary skill in the art that various embodiments may have fewer or a greater number of components or subcomponents than those illustrated in FIG. 2a. Thus, the depiction of the environment in FIG. 2a or in other figures should be taken as being illustrative in nature and not limiting to the scope of the disclosure.

In the illustrated embodiment, the device identification system 200A includes an analysis engine 202, evaluation engine 204 and an action engine 206. In some other embodiments, the device identification system 200A may include a subset or a superset of the illustrated components. For example, in an embodiment, the device identification system may include only the evaluation engine. In another embodiment, the device identification system may include only the analysis engine and the evaluation engine. In some embodiments, some or all of the components discussed herein may be combined or further divided into subcomponents. Some or all of the components may be implemented by the provider of the system or by a third party service provider.

The analysis engine 202 may be configured to generate, based on historical data 201, evaluation rules or rules 208 that may be used to determine the uniqueness of device identifiers. Specifically, the rules may be used, for example, by the evaluation engine 204, to derive a quality score or grade 210 associated with a given device identifier. Such rules may be derived based on historical data obtained from many user devices and many transactions. Various techniques may be used to derive the rules including machine learning techniques, neural networks, fuzzy logic, statistical analysis (e.g., logistical regression), and the like. Rules may be generated automatically with aid of a processor. Human intervention may or may not be required for generating the rules.

The historical data may include data (including statistics) related to past user activities, transactions, requests, responses, usage of computing resources and the like. In some cases, the historical data may include information indicative of reliability, trustworthiness or uniqueness of user devices. For example, the historical data may indicate that a certain IP address or a certain operating system is susceptible to security problems (e.g., virus, Denial of Service (DoS) attack, session hijacking, Man-in-the-Middle (MITM) or Man-in-the-Browser (MITB) attacks, etc.). As another example, the historical data may indicate that certain types of devices tend to share the same device identifiers.

The evaluation engine 204 may be configured to determine uniqueness of a device identifier based on evaluation rules 208, discussed above. To that end, the evaluation engine 204 may be configured to obtain device data 205 associated with a device. In some embodiments, the device data may include a plurality of parameters associated with or used to derive a device identifier. In some cases, device data may include the device identifier itself. The plurality of parameters may include any hardware and/or software parameters associated with a device identified by the device identified such user agent identifier, IP address, user identity information, user credentials, network protocols, service endpoint, service method, HTTP method and/or status code, operating system, locale or language code, processor architecture, device type (e.g., desktop, mobile phone, etc.) and the like. In some embodiments, the plurality of parameters may be associated with a particular transaction or network event.

Based at least in part on the plurality of parameters (e.g., device type), the evaluation engine 204 may be configured to select and apply some or all of the evaluation rules 208 made available by the analysis engine 202. In some embodiments, the evaluation rules may be stored in a data store or data file that is made available to the evaluation engine 204. The evaluation rules may be applied to at least some of the parameters to derive a device quality score or grade using methods discussed herein.

In some embodiments, the rules may be used to determine quality score and/or quality grade without requiring access to historical data. Such determination may be performed, for example, by the evaluation engine. Such a rules-only approach may be beneficial. For example, in some cases, the historical data may include sensitive or personally identifying information such as credit card information. In such cases, it may be undesirable to allow certain entities to have access to the historical data, for example, for privacy concerns.

The device quality score or grade may be used by the action engine 206 to determine an action 207. In various embodiments, the action engine may include or be included in one or more web servers, data servers, security and/or fraud detection servers and the like. The action may include retrieval, storage, processing, modification, transmission, or the like, of one or more responses to a request, internal or external messages or instructions, content data, and the like. In some cases, device identification system discussed herein may be used to detect fraudulent and/or malicious attacks such as session hijacking, MITM/MITB attacks, harvesting P2P networks, and the like. In some cases, device identification system may be used to determine suitable content (e.g., advertisement) to provide.

In some embodiments, analysis engine, the evaluation engine and the action engine may reside on the same or different computing devices and may each be implemented by one or more computing devices or processes. In some embodiments, the rules, the device quality scores or grades, and/or the actions may be generated in real or nearly real time as the data is coming in, or in an asynchronous fashion such as in using batch processing. In some embodiments, the generation of rules and the evaluation of the uniqueness of device identifiers can be independent from each other. The rules may be generated and/or updated at a different time schedule than that for the evaluation of the device identifiers. For example, in an embodiment, the rules are generated ahead of time and updated on a periodic basis. Independently or asynchronously to the generation and/or update of rules, device identifiers may be evaluated in real or nearly real time using the rules.

In some embodiments, analysis engine, the evaluation engine and the action engine may be configured to provide the various functionalities discussed herein in a synchronous or asynchronous fashion. For example, the generation of rules may be performed offline, in an asynchronous fashion. The evaluation of device quality score or grade may be performed in real time or nearly real time as the device data is received. The determining of a suitable action based on the device score and/or grade may be performed in real time or nearly real time.

FIG. 2b illustrates example components of a device identification system 200B, in accordance with another embodiment. In this example, the device identification system 200B includes a data collector 214 residing on a user or client device 212. The data collector may be implemented as a browser script using JavaScript or any other scripting language. The data collector may be configured to communicate with a device identification service 216. For example, the data collector may be configured to collect parameter information about the user device such as discussed herein and transmit such parameter information to the device identification service 216, for example, using an API provided by the device identification service. In some embodiments, the collection and/or communication with the device identification service may be triggered by an event such as a browser event. For example, the event may include a click on a portion (e.g., a button or a link) of a web page, loading of a web page and the like.

The device identification system 200B includes a device identification service 216 that may be implemented as a web service. The device identification service 216 may be implemented by one or more servers. The servers implementing the device identification service 216 may be owned and/or provided by a content or service provider for the user device (e.g., banking, ecommerce, retail) or by the provider of the device identification system 200B.

In some embodiments, the device identification service 216 may be configured to receive parameter information provided by the data collector of the user device and to provide a device identifier and/or device quality score or grade based on the parameter information. To that end, the device identification service 216 may utilize an evaluation engine 218. The evaluation engine 218 may be configured to calculate a device identifier and/or a device quality score or grade based on the parameter information. In some embodiments, the evaluation engine 218 may be implemented using one or more server-side library files.

In some embodiments, some or all of the parameters may be used to derive the device identifier. For example, the parameters may be concatenated, combined, appended, hashed, encrypted, and otherwise processed. For example, in an embodiment, the device identifier may include a 40-character SHA-1 hash.

In some embodiments, the device quality score or grade may be evaluated based on some or all of the parameters. For example, a lookup table (e.g., stored in memory) may be used to determine the weight values associated with some or all of the parameters. The weight values may or may not be further weighted, combined or otherwise processed to derive a final device quality score or grade. The device quality score may be categorized into a device quality grade. In some embodiments, the lookup table and the algorithm for deriving the quality score or grade may be included on one or more rules that are pre-determined based on historical data such as past transactions and/or user activities related to one or more websites or web services. Thus, access to the actual historical data may not be required for the evaluation of the quality scores or grades. In some embodiments, the generation of the device identifiers and/or the associated device quality scores and/or grades may be performed in real time or nearly real time with respect to the receipt of the parameter information. In other embodiments, any or all of the above operations may be performed in an asynchronous mode, for example, using batch processing.

In some embodiments, the generated device identifier and associated device quality score and/or grade may be stored in a data store 220. The data store 220 may include a user ID map (not shown) or a similar data structure configured to store a mapping between device identifiers and device quality scores and/or grades. In some embodiments, the data store 220 may include a memory of a server, one or more data storage device (e.g., SSD, hard disk, taps), or a cloud-based storage service such as discussed in connection with FIG. 1. The data store 220 may or may not be owned and/or operated by the same as the provider of the device identification service 216. For example, the user ID map may be stored at least in part on a customer server and/or a fraud-detection system.

In some embodiments, the storing of the device identifiers and/or the associated device quality scores and/or grades may be performed in real time or nearly real time as the above information is generated. In other embodiments, any or all of the above operations may be performed in an asynchronous mode, for example, using batch processing.

In various embodiments, the user ID map may be used by any suitable entity for any suitable purpose. For example, in an embodiment, the user ID map may be used by a content provider to determine the type of content to provide to a user device. More targeted content (e.g., advertisement) may be provided for device identifiers with higher quality grades and less targeted content may be provided for device identifiers with lower quality grades. In another embodiment, the user ID map may be used by a fraud detection system to detect and/or prevent online fraud.

In some embodiments, the user ID map may be used to update and/or refine the evaluation rules (e.g., including weight lookup table, device score computation algorithm) discussed herein. For example, the user ID map may be provided for research purposes. The research may be performed by a provider of the device identification system or a third party service provider.

FIG. 3 illustrates example components of a computer device 300 for implementing aspects of the present invention, in accordance with an embodiment. In another embodiment, the computer device 300 may be configured to implement a user device such as a user device 102 discussed in connection with FIG. 1 and/or components or aspects of the device identification system such as described in connection with FIGS. 1 and 2. In some embodiments, computing device 300 may include many more components than those shown in FIG. 3. However, it is not necessary that all of these components be shown in order to disclose an illustrative embodiment.

As shown in FIG. 3, computing device 300 includes a network interface 302 for connecting to a network such as discussed above. In various embodiments, the computing device 300 may include one or more network interfaces 302 for communicating with one or more types of networks such as the Internet, wireless networks, cellular networks, and any other network.

In an embodiment, computing device 300 also includes one or more processing units 304, a memory 306, and an optional display 308, all interconnected along with the network interface 302 via a bus 310. The processing unit(s) 304 may be capable of executing one or more methods or routines stored in the memory 306. The display 308 may be configured to provide a graphical user interface to a user operating the computing device 300 for receiving user input, displaying output, and/or executing applications. In some cases, such as when the computing device 300 is a server, the display 308 may be optional.

The memory 306 may generally comprise a random access memory (“RAM”), a read only memory (“ROM”), and/or a permanent mass storage device, such as a disk drive. The memory 306 may store program code for an operating system 312, one or more device identification routines 314, and other routines. In various embodiments, the program code may be stored on a computer-readable storage medium, for example, in the form of a computer program comprising a plurality of instructions executable by one or more processors. The computer-readable storage medium may be non-transitory. The one or more device identification routines 314, when executed, may provide various functionalities associated with the device identification system as described herein.

In some embodiments, the software components discussed above may be loaded into memory 306 using a drive mechanism associated with a non-transient computer readable storage medium 318, such as a floppy disc, tape, DVD/CD-ROM drive, memory card, USB flash drive, solid state drive (SSD) or the like. In other embodiments, the software components may alternatively be loaded via the network interface 302, rather than via a non-transient computer readable storage medium 318. In an embodiment, the computing device 300 also include an optional time keeping device (not shown) for keeping track of the timing of transactions or network events.

In some embodiments, the computing device 300 also communicates via bus 310 with one or more local or remote databases or data stores such as an online data storage system via the bus 310 or the network interface 302. The bus 310 may comprise a storage area network (“SAN”), a high-speed serial bus, and/or via other suitable communication technology. In some embodiments, such databases or data stores may be integrated as part of the computing device 300.

FIG. 4 illustrates an example process 400 for implementing the present invention, in accordance with an embodiment. Aspects of the process 400 may be performed, for example, by a device identification system such as discussed in connection with FIGS. 1 and 2 or one or more computing devices such as discussed in connection with FIG. 3. Some or all aspects of the process 400 (or any other processes described herein, or variations and/or combinations thereof) may be performed under the control of one or more computer/control systems configured with executable instructions and may be implemented as code (e.g., executable instructions, one or more computer programs or one or more applications) executing collectively on one or more processors, by hardware or combinations thereof. The code may be stored on a computer-readable storage medium, for example, in the form of a computer program comprising a plurality of instructions executable by one or more processors. The computer-readable storage medium may be non-transitory. The order in which the operations are described is not intended to be construed as a limitation, and any number of the described operations may be combined in any order and/or in parallel to implement the processes.

In an embodiment, the process 400 includes obtaining 402 a set of rules based on historical data. In various embodiments, the rules may include the evaluation rules, discussed herein, that may be used to determine the uniqueness of a device identifier. For example, the rules may include one or more parameter maps that map parameter values (original or derived) to weight values. The rules may further include formulas, algorithms, and the like for using the maps to (e.g., combining the weight values) to derive a device quality score and/or device quality grade. Such a device quality score or grade may be indicative of the uniqueness of the device identifier. In a typical embodiment, the size of the rules is a fraction of the amount of the historical data based on which the rules are derived.

As discussed above, historical data may include any data related to past data transactions, user activities, usage of computing and network resources and the like. In some cases, the historical data may include information indicative of reliability, trustworthiness or uniqueness of user devices or device identifiers. In some embodiments, the historical data may be obtained from a third-party data or service provider and/or accumulated by a provider of the device identification system. For example, historical data may include interactions with content providers, ecommerce or online retail service providers, banking, credit card, or financial service providers, airlines, travel service providers, and the like.

In some embodiments, the rules may be generated using a variety of machine learning and/or data mining techniques such as statistical analysis, neural networks, and the like. In one embodiment, some of the rules may be defined or specified by humans. In some embodiments, some of the rules may be generated from scratch or provided by a third-party provider.

In an embodiment, the process 400 includes determining 404 the uniqueness of a device identifier based on the rules discussed above and a plurality of parameters associated with the device identifier. In some embodiments, a device identifier may be used to identify the source or origin or a transaction, request, or network event. In some embodiments, a device identifier may be determined and/or derived based on any one or combination of one or more parameters such as described herein. For example, the device identifier may be based on one or more hardware and/or software settings or attributes of a device. For example, a device identifier may include or be based on an IP address associated with an HTTP request. As another example, a device identifier may include or be based on a username associated with an online account and a user agent identifier. The device identifier may include a device fingerprint without regard to user information. In one embodiment, the device identifier may be independent from the parameters described herein. In a typical embodiment, a device identifier is obtained or derived without the awareness of the originator of the transaction or network event identified by the device identifier. In other embodiments, the device identifier may be obtained with user awareness.

In various embodiments, a plurality of parameters such as those discussed herein may be obtained in connection with the device identifier. Such parameters may be obtained from the transaction or network event identified by the device identifier. Such parameters may be obtained, for example, by analyzing the metadata and/or data associated with a request, parsing a network log file, utilizing any suitable web analytics tools, and the like. In a typical embodiment, such parameters are obtained without user awareness. In some embodiments, such parameters may be collected with user awareness. In some embodiments, the device identifier and/or parameters may be obtained without downloading anything to the device (i.e., using a tag-free technique) or by downloading something (e.g., a cookie or browser script) to the device.

Based on the rules and the plurality of parameters, uniqueness of the device identifier may be determined. In some embodiments, the uniqueness of the device identifier may be represented by a device quality score or grade discussed herein. In other embodiments, the uniqueness of the device identifier may be represented by any other suitable representations. More details for determining the uniqueness of a device identifier are discussed below in connection with FIGS. 5-7.

In an embodiment, the process 400 includes determining 406 a suitable action or measure to take based on the uniqueness of the device identifier. Such action may be selected among a plurality of actions based on a determined device quality score or quality grade for the device identifier. For example, different action(s) may be taken if the device identifier is more unique than if the device identifier is less unique, or if the quality score or grade is different. The action may include an active action such as the retrieval, storage, processing, modification, transmission, or the like, of one or more responses, messages, instructions, and the like. In an embodiment, the action may include not doing something. In some embodiments, determining the suitable action may include comparing the device quality score or grade with a predefined threshold value and selecting the suitable action based on the result of the comparison. For example, if a device identifier is determined to be more likely to be unique (e.g., having a quality score or grade higher than a predefined threshold value), then a more targeted advertisement may be provided. Conversely, if a device identifier is determined to be less likely to be unique (e.g., having a quality score or grade equal or less than the predefined threshold value), then a less targeted advertisement may be provided. Similarly, a more severe or drastic security or anti-fraud measure (e.g., adding the device identifier to a blacklist) may be taken if a device identifier is determined to be more likely to be unique. On the other hand, a more moderate security measure may be taken if a device identifier is determined to be less likely to be unique.

In some embodiments, step 402 may be performed on a periodic basis (e.g., daily, weekly, monthly). In some embodiments, steps 404 and 406 may be performed for each of a plurality of transactions in real or nearly real time or in an asynchronous fashion (i.e., not in real or nearly real time).

FIG. 5 illustrates an example process 500 for determining the uniqueness of a device identifier, in accordance with an embodiment. Aspects of the process 500 may be performed, for example, by the evaluation engine discussed in connection with FIG. 2a or 2b.

In an embodiment, the process 500 includes obtaining 502 a set of evaluation rules based on historical data. As discussed above, such rules may be made available via a data file, data storage system, web service, or any other suitable interface. In various embodiments, the rules may be made available via the push or pull technologies or a combination of both. In some embodiments, once the set of rules are obtained, they can be used to evaluate the uniqueness of one or more (e.g., hundreds or thousands of) device identifiers. In some embodiments, the rules may be updated occasionally (e.g., on a periodic basis).

In an embodiment, the process 500 includes obtaining 404 a plurality of parameters associated with a device identifier. The device identifier may be associated with a device, a group of devices, a transaction, a user, an organization or any other entity. The plurality of parameters may be obtained from the device identifier itself, from the entity associated with the device identifier, from log files, from real-time analysis of network traffic, or from other channels using any suitable method.

Based on the rules and the plurality of parameters, the uniqueness of the device identifier may be determined 506, for example, using the process discussed below in connection with FIGS. 6-7.

FIG. 6 illustrates an example process 600 for determining the uniqueness of a device identifier, in accordance with an embodiment. Aspects of the process 600 may be performed, for example, by the evaluation engine discussed in connection with FIG. 2a or 2b.

In an embodiment, the process 600 includes obtaining 602 a plurality of parameters associated with a device identifier. In some embodiments, step 602 may be similar to step 504 discussed in connection with process 500 of FIG. 5.

In an embodiment, the process 600 includes selecting 604 applicable rules based on the plurality of parameters. The selection may be based on one, two or more of the plurality of parameters obtained above. For example, different sets of rules may be applicable to different sets of parameters. As another example, the applicable set of rules may be dictated by a subset of the plurality of parameters such as whether the device identifier is associated with a desktop or a mobile device, the operating system or processor architecture associated with the device identifier and the like.

In an embodiment, the process 600 includes applying the selected rules to determine 606 a plurality of weight values respectively associated with at least some of the plurality of parameters. For example, for each parameter value associated with a given parameter, a value-weight map may be used to look up a corresponding weight value. The value-weight map may be stored as a lookup table or similar data structure in memory or in another data storage medium. The value-weight map(s) may or may not be part of the rules obtained in step 604. The weight values may be used to obtain 608 a device quality score. In some embodiments, weight values may be further weighted before being combined to derive the device quality score. For example, the device quality score may include a linear combination of the weighted weight values. The formula or algorithm for combining the weight values may or may not be part of the rules obtained in step 604. The quality score may be represented by numeric value. The numeric value may fall within a predetermined numerical range. In general, a higher quality score indicates a higher likelihood that a device identifier is unique and vice versa.

In some embodiments, the quality score may be used to derive 610 a quality grade, such as grade A, B, C and so on. For example, a quality score between 80 and 100 may be categorized as quality grade A, a quality score between 60 and 80 may be categorized as quality grade B, and so on. The quality grade may be represented by numeric or non-numeric values. In some embodiments, the step 610 of deriving a quality grade may be optional.

FIG. 7 illustrates an example process 700 for calculating a device quality score, in accordance with an embodiment. The process 700 may be similar to the process 600 described above in connection with FIG. 6.

The illustrated process may be used to calculate the quality score associated with a transaction or network event 702 that is identified by a device identifier (not shown). A plurality of parameters may be associated with the network event, such as user agent identifier 704, IP address 706, language code 708, and the like. For each of the plurality of parameters, a weight value may be obtained based on a corresponding value-weight map. For example, the value-weight maps 716, 718, 720 may correspond respectively to the parameters 704, 706 and 708. In some embodiments, the parameter values may be transformed or otherwise used to obtain the parameter value used to look up the value-weight map, for example, according to parameter-specific usage rules 710, 712 and 714. For example, the length of the character string of a user agent identifier is to be calculated and used as the parameter value to look up the user agent value-weight map 716, according to rule 710. For example, a user agent identifier “UA_2.8.1” has a character length of 8 and a weight value of 0.5 according to the user agent value-weight map 716. For IP addresses, no transformation may be necessary and the original IP address may be used to look up the IP address value-weight map 618, according to rule 712. For example, an IP address “310.31.345” has an exact match in the IP address value-weight map 718 with a weight value of 0.4. And for language codes, the number of word count may be used to look up the language value-weight map 720, according to rule 714. For example, a language “Fr-fr” has a word count of 2 and a weight value of 0.02 according to the language value-weight map 720.

Once the weight values are obtained, they may be combined to derive the device quality score, for example, according to a formula and/or algorithm 722 to derive the final quality score 724. As discussed above, the weight values may be further weighted (e.g., according to the relative significance of the parameters) before they are combined. In some embodiments, the quality score may be normalized, for example, using a coefficient. Any suitable methods of normalization may be used to normalize the quality score.

In various embodiments, the usage rules, value-weight maps, formula or algorithm discussed above may collectively comprise the evaluation rules discussed herein. Such evaluation rules may be derived once based on historical data and used repeatedly to quickly calculate quality scores for many device identifiers. The evaluation rules may be updated, for example, on a periodic basis, as the historical data evolves.

FIG. 8 illustrates an example process 800 for determining the uniqueness of a device identifier, in accordance with an embodiment. Aspects of the process 800 may be performed, for example, by the device identification system discussed in connection with FIG. 2a or 2b.

In an embodiment, process 800 includes detecting 802 a network event such as a browser event (e.g., clicking of a control, scrolling, resizing, loading or closing of a web page, and the like). Such detection may be implemented by a browser script (e.g., JavaScript).

In an embodiment, in response to the detected network event, a plurality of parameters related to the device associated with the network event is obtained 804. Based on some or all of the obtained plurality of parameters, a device identifier and a device quality grade may be determined 806. Such determination may be further based on pre-calculated rules derived from historical data. The rules may be embodied by the value-weight map/lookup table discussed herein. Finally, the device identifier and the device quality grade may be stored 808, such as in a data store 220 discussed in connection with FIG. 2b. In some embodiments, storage 808 step may be optional and the device identifier and device quality grade may be used directly without being stored first.

In some embodiments, the present invention separates the calculation of rules based on historical data from the use of the rules to derive device identifier and/or device quality score or grade associated with the device identifier. Such separation serves to limit the access to the historical data alleviating concerns with respect to the privacy of the historical data. The separation also allows fast, real-time or nearly real time, and scalable determination of device identifiers and/or device quality scores and/or grades. Additionally, in some embodiments, the rules may be stored at a central location, making it easy to maintain and update the rules independently from the uses of the rules.

While preferred embodiments of the present invention have been shown and described herein, it will be obvious to those skilled in the art that such embodiments are provided by way of example only. Numerous variations, changes, and substitutions will now occur to those skilled in the art without departing from the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in practicing the invention. It is intended that the following claims define the scope of the invention and that methods and structures within the scope of these claims and their equivalents be covered thereby.

Claims

1. A computer-implemented method for automated selection of an electronic security action based on a uniqueness assessment of a device identifier, said computer-implemented method under the control of one or more computer systems configured with executable instructions and comprising: establishing an electronic connection with one or more databases storing at least: a plurality of evaluation rules; anda predetermined threshold associated with a range of quality scores, wherein the predetermined threshold is associated with a predetermined threshold value;detecting a first online session between a computer system and a first user device;detecting a network event during the first online session;receiving device data associated with the first user device, wherein the device data is captured during the first online session and comprises a plurality of parameters comprising a first parameter and a second parameter, and wherein the plurality of parameters are associated with the network event;determining a first device identifier associated with the first user device based at least in part on the plurality of parameters;receiving a first plurality of evaluation rules from the one or more databases, wherein the first plurality of evaluation rules are selected based at least in part on a value of the first parameter;determining, based at least in part on the first plurality of evaluation rules and values of the plurality of parameters, a plurality of weight values, wherein the plurality of weight values are associated with the plurality of parameters;calculating, based at least in part on the plurality of weight values, a first quality score associated with the first device identifier;receiving an electronic indication that the first online session is likely fraudulent;accessing the predetermined threshold from the one or more databases;comparing the first quality score with the predetermined threshold value;determining that the first quality score satisfies the predetermined threshold, wherein the determination that the first quality score satisfies the predetermined threshold indicates that the first device identifier is not shared by multiple devices; andbased at least in part on the determination that the first quality score satisfies the predetermined threshold, adding the first device identifier to a blacklist such that network traffic from any devices with an association to the first device identifier are blocked.

2. The computer-implemented method of claim 1, wherein the plurality of evaluation rules are generated using a machine learning technique, wherein the machine learning technique comprises a neutral network or a statistical analysis of historical data comprising device fingerprints and indication of whether such fingerprints uniquely identify a device.

3. The computer-implemented method of claim 2, wherein the calculating the first quality score is performed without access to the historical data.

4. The computer-implemented method of claim 2, wherein the plurality of evaluation rules is updated periodically based at least in part on changes of the historical data.

5. The computer-implemented method of claim 1, wherein: the plurality of weight values comprises: a first weight value associated with the first parameter; anda second weight value associated with the second parameter;the first weight value is determined based at least in part on a first weight-value map and the value of the first parameter; andthe second weight value is determined based at least in part on a second weight-value map and a value of the second parameter.

6. The computer-implemented method of claim 5, wherein: the plurality of evaluation rules comprises a first evaluation rule and a second evaluation rule;the first evaluation rule is associated with the first parameter and the first weight-value map; andthe second evaluation rule is associated with the second parameter and the second weight-value map.

7. The computer-implemented method of claim 1 further comprising: based at least in part on the determination that the first quality score satisfies the predetermined threshold, transmitting a targeted content to the devices with an association to the first device identifier, wherein the targeted content is customized based at least in part on the device data associated with the first user device.

8. The computer-implemented method of claim 1 further comprising: determining that the first quality score does not satisfy the predetermined threshold, wherein the determination that the first quality score does not satisfy the predetermined threshold indicates that the first device identifier is shared by multiple devices; andbased at least in part on the determination that the first quality score does not satisfy the predetermined threshold, monitoring network traffic from the devices with an association to the first device identifier.

9. A computer system for automatically selecting an electronic security action based on a uniqueness assessment of a device identifier, the system comprising: one or more processors; anda memory, including instructions executable by the one or more processors to cause the computer system to at least: establish an electronic connection with one or more database storing at least: a plurality of evaluation rules; anda predetermined threshold associated with a range of quality scores, wherein the predetermined threshold is associated with a predetermined threshold value;detect a first online session between a computer system and a first user device;detect a network event during the first online session;receive device data associated with the first user device, wherein the device data is captured during the first online session and comprises a plurality of parameters comprising a first parameter and a second parameter, and wherein the plurality of parameters are associated with the network event;determine a first device identifier associated with the first user device based at least in part on the plurality of parameters;receive a first plurality of evaluation rules from the one or more databases, wherein the first plurality of evaluation rules are selected based at least in part on a value of the first parameter;determine, based at least in part on the first plurality of evaluation rules and values of the plurality of parameters, a plurality of weight values, wherein the plurality of weight values are associated with the plurality of parameters;calculate, based at least in part on the plurality of weight values, a first quality score associated with the first device identifier;receive an electronic indication that the first online session is likely fraudulent;access the predetermined threshold from the one or more databases;compare the first quality score with the predetermined threshold value;determine that the first quality score satisfies the predetermined threshold, wherein the determination that the first quality score satisfies the predetermined threshold indicates that the first device identifier is not shared by multiple devices; andbased at least in part on the determination that the first quality score satisfies the predetermined threshold, add the first device identifier to a blacklist such that network traffic from any devices with an association to the first device identifier are blocked.

10. The system of claim 9, wherein the plurality of evaluation rules are generated using a machine learning technique, wherein the machine learning technique comprises a neutral network or a statistical analysis of historical data comprising device fingerprints and indication of whether such fingerprints uniquely identify a device.

11. The system of claim 10, wherein the plurality of evaluation rules is updated periodically based at least in part on changes of the historical data.

12. The system of claim 10, wherein the first quality score is calculated without access to the historical data.

13. The system of claim 9, wherein the instructions executable by the one or more processors further cause the computer system to: based at least in part on the determination that the first quality score satisfies the predetermined threshold, transmit a targeted content to the devices with an association to the first device identifier, wherein the targeted content is customized based at least in part on the device data associated with the first user device.

14. The system of claim 9, the instructions executable by the one or more processors further cause the computer system to: determine that the first quality score does not satisfy the predetermined threshold, wherein the determination that the first quality score does not satisfy the predetermined threshold indicates that the first device identifier is shared by multiple devices; andbased at least in part on the determination that the first quality score does not satisfy the predetermined threshold, monitor network traffic from the devices with an association to the first device identifier.

15. A non-transitory computer storage having stored thereon a computer program, the computer program including executable instructions that instruct a computer system to at least: establish an electronic connection with one or more databases storing at least: a plurality of evaluation rules; anda predetermined threshold associated with a range of quality scores, wherein the predetermined threshold is associated with a predetermined threshold value;detect a first online session between a computer system and a first user device;detect a network event during the first online session;receive device data associated with the first user device, wherein the device data is captured during the first online session and comprises a plurality of parameters comprising a first parameter and a second parameter, and wherein the plurality of parameters are associated with the network event;determine a first device identifier associated with the first user device based at least in part on the plurality of parameters;receive a first plurality of evaluation rules from the one or more databases, wherein the first plurality of evaluation rules are selected based at least in part on a value of the first parameter;determine, based at least in part on the first plurality of evaluation rules and values of the plurality of parameters, a plurality of weight values, wherein the plurality of weight values are associated with the plurality of parameters;calculate, based at least in part on the plurality of weight values, a first quality score associated with the first device identifier;receive an electronic indication that the first online session is likely fraudulent;access the predetermined threshold from the one or more databases;compare the first quality score with the predetermined threshold value;determine that the first quality score satisfies the predetermined threshold, wherein the determination that the first quality score satisfies the predetermined threshold indicates that the first device identifier is not shared by multiple devices; andbased at least in part on the determination that the first quality score satisfies the predetermined threshold, add the first device identifier to a blacklist such that network traffic from any devices with an association to the first device identifier are blocked.

16. The non-transitory computer storage of claim 15, wherein the plurality of evaluation rules are generated using a machine learning technique, wherein the machine learning technique comprises a neutral network or a statistical analysis of historical data comprising device fingerprints and indication of whether such fingerprints uniquely identify a device.

17. The non-transitory computer storage of claim 16, wherein the plurality of evaluation rules is updated periodically based at least in part on changes of the historical data.

18. The non-transitory computer storage of claim 16, wherein the first quality score is calculated without access to the historical data.

19. The non-transitory computer storage of claim 15, wherein the executable instructions further instruct the computer system to: based at least in part on the determination that the first quality score satisfies the predetermined threshold, transmit a targeted content to the devices with an association to the first device identifier, wherein the targeted content is customized based at least in part on the device data associated with the first user device.

20. The non-transitory computer storage of claim 15, wherein the executable instructions further instruct the computer system to: determine that the first quality score does not satisfy the predetermined threshold, wherein the determination that the first quality score does not satisfy the predetermined threshold indicates that the first device identifier is shared by multiple devices; andbased at least in part on the determination that the first quality score does not satisfy the predetermined threshold, monitor network traffic from the devices with an association to the first device identifier.