Subject type determination method and apparatus, computing device, and storage medium

Abstract

The disclosure provides a subject type determination method, characterized in that the method comprises: obtaining resource transaction information of a target subject in a target time period, the resource transaction information comprising resource transfer-in information, resource transfer-out information and resource retention information related to the target subject; determining, based on the resource transaction information, a resource transfer-in feature value corresponding to the resource transfer-in information, a resource transfer-out feature value corresponding to the resource transfer-out information, and a resource retention feature value corresponding to the resource retention information, wherein the resource transfer-in feature value, the resource transfer-out feature value, and the resource retention feature value respectively represent the degree of change in resource transaction represented by the corresponding resource transaction information of the target subject in the target time period; and determining the subject type of the target subject based on the resource transfer-in feature value, the resource transfer-out feature value, and the resource retention feature value.

Description

Technical Field

[0001] This disclosure relates to the technical field of classification, specifically to a method and apparatus for determining subject type, as well as corresponding computing devices, computer-readable storage media, and computer program products. Background Technology

[0002] Different types of entities tend to perform different types of operations. Depending on the nature of the operation, different types of operations should be treated differently—for example, allowing the operation to proceed or preventing it from proceeding. Similarly, different types of entities should also be treated differently—for example, allowing one type of entity to perform a certain operation while disallowing another type of entity from performing that operation.

[0003] In related technologies, the determination of an entity's type is often based on static information associated with that entity. For example, when an entity is an organization engaged in a certain operational activity, static information related to that entity (such as the industry in which the organization operates, its operating model, and its annual operating reports) can be used to determine its entity type. Based on this determined entity type, certain operations can be permitted or prohibited (e.g., allowing or prohibiting the transfer of resources to that entity). However, an entity's resource trading activities are often dynamic, and entities engaging in abnormal resource trading may have static information similar to other entities engaging in normal resource trading. This makes it difficult to use related technologies to identify these entities engaging in abnormal resource trading in a timely and accurate manner, and consequently, to take appropriate measures against them in a timely manner. Summary of the Invention

[0004] In view of this, the present disclosure provides a method and apparatus for determining subject type, a computing device, a computer-readable storage medium, and a computer program product to alleviate, mitigate, or even eliminate the above-mentioned problems.

[0005] According to one aspect of this disclosure, a method for determining the type of an entity is provided. The method includes: acquiring resource transaction information of a target entity within a target time period, the resource transaction information including resource transfer-in information, resource transfer-out information, and resource retention information related to the target entity; determining, based on the resource transaction information, resource transfer-in characteristic values ​​corresponding to the resource transfer-in information, resource transfer-out characteristic values ​​corresponding to the resource transfer-out information, and resource retention characteristic values ​​corresponding to the resource retention information, wherein the resource transfer-in characteristic values, resource transfer-out characteristic values, and resource retention characteristic values ​​are respectively used to represent the degree of change in resource transactions represented by the corresponding resource transaction information of the target entity within the target time period; and determining the entity type of the target entity based on the resource transfer-in characteristic values, resource transfer-out characteristic values, and resource retention characteristic values.

[0006] According to some embodiments of this disclosure, determining the resource transfer-in feature value corresponding to the resource transfer-in information, the resource transfer-out feature value corresponding to the resource transfer-out information, and the resource retention feature value corresponding to the resource retention information based on the resource transaction information includes: determining the first multi-scale permutation entropy corresponding to the resource transfer-in information, the second multi-scale permutation entropy corresponding to the resource transfer-out information, and the third multi-scale permutation entropy corresponding to the resource retention information based on the resource transaction information; and using the first multi-scale permutation entropy as the resource transfer-in feature value, the second multi-scale permutation entropy as the resource transfer-out feature value, and the third multi-scale permutation entropy as the resource retention feature value.

[0007] According to some embodiments of this disclosure, determining the first multi-scale permutation entropy corresponding to the resource transfer-in information, the second multi-scale permutation entropy corresponding to the resource transfer-out information, and the third multi-scale permutation entropy corresponding to the resource retention information based on the resource transaction information includes: treating the resource transfer-in information, the resource transfer-out information, and the resource retention information as resource processing information, respectively; and using a parameterized multi-scale permutation entropy model corresponding to the resource processing information to determine the first multi-scale permutation entropy, the second multi-scale permutation entropy, and the third multi-scale permutation entropy based on the resource processing information, wherein the parameters of the multi-scale permutation entropy model include signal length, embedding dimension, delay, and scale factor, and includes: coarsening the transaction sequence corresponding to the resource processing information according to the scale factor to obtain multiple coarse-grained sequences at different scales corresponding to the transaction sequence, wherein the length of the transaction sequence is... The length is the same as the signal length, and the number of sequence elements contained in each of the plurality of coarse-grained sequences is the same as the embedding dimension; phase space reconstruction is performed on each of the plurality of coarse-grained sequences to obtain a reconstruction matrix corresponding to each coarse-grained sequence, wherein the difference between two adjacent matrix elements in any row of the reconstruction matrix and the two sequence indices in the coarse-grained sequence corresponding to the reconstruction matrix is ​​the same as the delay amount; for each of the plurality of coarse-grained sequences, the matrix elements in each row of the reconstruction matrix corresponding to each coarse-grained sequence are arranged in ascending order to obtain the corresponding ascending sequence, and the permutation entropy of the corresponding coarse-grained sequence is determined according to the frequency of the corresponding ascending sequence in the corresponding reconstruction matrix; and the multi-scale permutation entropy corresponding to the resource processing information is determined based on the permutation entropy of each of the plurality of coarse-grained sequences.

[0008] According to some embodiments of this disclosure, the values ​​of the parameters of the multi-scale permutation entropy model are determined through the following steps: generating a particle swarm corresponding to the parameters of the multi-scale permutation entropy model, wherein each particle in the particle swarm has an initial velocity and an initial position; based on the initial velocity and initial position of each particle in the particle swarm, iteratively executing the following steps to reach a preset termination condition to obtain the updated position of each particle: calculating the value of the objective function of the corresponding particle in the current iteration based on the position of each particle in the current iteration; comparing the value of the objective function of each particle in the current iteration with the value of the objective function of the corresponding particle in the previous iteration to determine the individual optimal position of each particle; determining the group optimal position of the particle swarm based on the individual optimal position of each particle; updating the velocity and position of the corresponding particle based on the individual optimal position of each particle and the group optimal position of the particle swarm; and determining the values ​​of the parameters of the multi-scale permutation entropy model based on the updated positions of each particle in the particle swarm.

[0009] According to some embodiments of this disclosure, the preset termination condition includes one of the following conditions: the number of iterations is reached; and the difference between the objective function value of each particle in the current iteration and the objective function value of the corresponding particle in the previous iteration is less than a first preset threshold.

[0010] According to some embodiments of this disclosure, determining the subject type of the target subject based on the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value includes: determining a first type, a second type, and a third type of the target subject based on the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value, respectively; and determining the subject type of the target subject based on the first type, the second type, and the third type.

[0011] According to some embodiments of this disclosure, determining the first type, second type, and third type of the target entity based on the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value includes: using the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value as resource transaction characteristic values, and using a trained entity classification model corresponding to the resource transaction characteristic values ​​to determine the first type, second type, and third type of the target entity based on the resource transaction characteristic values; wherein, the trained entity classification model is trained through the following steps: determining the sample transaction corresponding to each sample entity based on multiple sample resource transaction information of multiple sample entities. The feature value is used to represent the degree of change in resource transactions represented by the corresponding resource transaction information of the corresponding sample subject within the corresponding sample time period. The feature value of each sample subject is input into the subject classification model to obtain the predicted subject type for each sample subject. Based on the predicted subject type and the sample subject type, a loss function is determined. The parameters of the subject classification model are adjusted to minimize the loss function.

[0012] According to some embodiments of this disclosure, the sample subject type includes a first sample type and a second sample type, wherein the degree of change in resource transactions of the sample subject corresponding to the first sample type within the corresponding sample time period does not exceed a second preset threshold, and the degree of change in resource transactions of the sample subject corresponding to the second sample type within the corresponding sample time period exceeds the second preset threshold.

[0013] According to another aspect of this disclosure, a subject type determination apparatus is provided, comprising: a resource transaction information acquisition module configured to acquire resource transaction information of a target subject within a target time period, the resource transaction information including resource transfer-in information, resource transfer-out information, and resource retention information related to the target subject; a transaction feature value determination module configured to determine, based on the resource transaction information, a resource transfer-in feature value corresponding to the resource transfer-in information, a resource transfer-out feature value corresponding to the resource transfer-out information, and a resource retention feature value corresponding to the resource retention information, wherein the resource transfer-in feature value, the resource transfer-out feature value, and the resource retention feature value are respectively used to represent the degree of change in resource transactions represented by the corresponding resource transaction information of the target subject within the target time period; and a subject type determination module configured to determine the subject type of the target subject based on the resource transfer-in feature value, the resource transfer-out feature value, and the resource retention feature value.

[0014] According to another aspect of this disclosure, a computing device is provided, comprising: a memory configured to store computer-executable instructions; and a processor configured to perform any of the methods provided according to the foregoing aspects of this disclosure when the computer-executable instructions are executed by the processor.

[0015] According to another aspect of this disclosure, a computer-readable storage medium is provided that stores computer-executable instructions that, when executed, perform any of the methods provided according to the foregoing aspects of this disclosure.

[0016] According to yet another aspect of this disclosure, a computer program product is provided, including computer-executable instructions that, when executed by a processor, perform any of the methods provided according to the foregoing aspects of this disclosure.

[0017] According to the subject type determination method provided in this disclosure, the resource transfer-in characteristic value corresponding to the resource transfer-in information, the resource transfer-out characteristic value corresponding to the resource transfer-out information, and the resource retention characteristic value corresponding to the resource retention information can be determined based on the resource transaction information of the target subject within a target time period. The resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value are respectively used to represent the degree of change in resource transactions represented by the corresponding resource transaction information of the target subject within the target time period. Therefore, the dynamic transaction characteristics of the target subject within the target time period can be quantified by the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value, enabling the determination of the subject type of the target subject based on the transaction characteristic values. This helps to identify subjects engaging in abnormal resource transactions in a timely and accurate manner, and thus helps to take appropriate measures against these subjects.

[0018] These and other aspects of this disclosure will be clear from the embodiments described below, and will be elucidated with reference to the embodiments described below. Attached Figure Description

[0019] In the following description of exemplary embodiments in conjunction with the accompanying drawings, further details, features, and advantages of the present disclosure are disclosed, in which:

[0020] Figure 1 The illustrations illustrate example scenarios in which the technical solutions provided according to some embodiments of this disclosure can be applied;

[0021] Figure 2 An example flowchart illustrating a method for determining a subject type according to some embodiments of the present disclosure is shown schematically;

[0022] Figure 3 An example flowchart illustrating a method for determining multi-scale permutation entropy corresponding to resource transaction information according to some embodiments of the present disclosure is shown.

[0023] Figure 4 An example flowchart illustrating a method for determining the values ​​of parameters of a multi-scale permutation entropy model according to some embodiments of the present disclosure is shown schematically.

[0024] Figure 5 An example flowchart illustrating a method for determining transaction characteristic values ​​of a target entity based on resource transaction information according to some embodiments of the present disclosure is shown.

[0025] Figure 6 An example flowchart illustrating a method for determining the subject type of a target subject based on transaction feature values ​​according to some embodiments of the present disclosure is shown.

[0026] Figure 7 An example flowchart illustrating another method for determining the subject type of a target subject based on transaction feature values ​​according to some embodiments of this disclosure is shown.

[0027] Figure 8 An example block diagram of a subject type determination apparatus according to some embodiments of the present disclosure is shown schematically;

[0028] Figure 9 An example system is illustrated, which includes an example computing device representing one or more systems and / or devices that can implement the various technologies described herein. Detailed Implementation

[0029] Several embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings to enable those skilled in the art to implement the technical solutions of this disclosure. The technical solutions of this disclosure can be embodied in many different forms and for many different purposes, and should not be limited to the embodiments set forth herein. These embodiments are provided to make the technical solutions of this disclosure clear and complete, but the described embodiments do not limit the scope of protection of this disclosure.

[0030] Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having meanings consistent with their meanings in the relevant field and / or the context of this specification, and will not be interpreted in an idealized or overly formal sense unless expressly defined herein.

[0031] Before detailing the embodiments of this disclosure, the relevant concepts will first be explained.

[0032] Machine Learning (ML): ML is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory, and many other disciplines. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instructional learning. Common machine learning models (e.g., fully connected neural network models, support vector machine models, etc.) can be applied to the methods provided in this disclosure to determine the subject type of the target agent.

[0033] Figure 1 The illustration depicts an example scenario 100 where the technical solutions provided according to some embodiments of this disclosure can be applied. For example... Figure 1 As shown, scenario 100 may include user 110, terminal device 120 (e.g., computer), terminal device 130 (e.g., tablet), network 140, and remote facility 150. As an example, remote facility 150 includes server 151 and optionally also includes database device 152 for storing resource transaction information (e.g., resource transaction information of a target subject within a target time period), and these servers or devices can communicate via network 140.

[0034] In this disclosure, the target entity may be an institution, organization, or individual that aims to provide certain services or goods or engages in certain operational activities, or the account used by such institution, organization, or individual to conduct resource transactions. The machinery and equipment used for conducting resource transactions include, but are not limited to, the institution's, organization's, or individual's computer, server, or mobile device such as a mobile phone.

[0035] An application (not shown) may be installed on terminal device 120 or 130. This application provides a visual user interface for user 110, who interacts with the application on this user interface to perform resource trading operations. For example, during a target time period, user 110 operates an entity (e.g., a funds account) through terminal device 120 or 130 to perform resource trading operations with another entity (e.g., a target entity). The resource trading operation may be a resource transfer operation to transfer resources (e.g., funds) to the target entity (e.g., a funds account), or a resource transfer operation to transfer resources (e.g., funds) out of the target entity (e.g., a funds account). Server 151 may obtain relevant resource trading information from the terminal devices (e.g., terminal device 120, terminal device 130) performing these resource trading operations via network 140. Alternatively, server 151 can obtain resource transaction information of the target entity within a target time period from a storage device (e.g., database device 152) storing corresponding resource transaction information. This resource transaction information includes resource transfer-in information, resource transfer-out information, and resource retention information related to the target entity. After obtaining the resource transaction information of the target entity within the target time period, server 151 can determine the resource transfer-in feature value corresponding to the resource transfer-in information, the resource transfer-out feature value corresponding to the resource transfer-out information, and the resource retention feature value corresponding to the resource retention information based on the resource transaction information (e.g., through a trained feature extraction model). The resource transfer-in feature value, the resource transfer-out feature value, and the resource retention feature value... The resource retention characteristic values ​​are used to represent the degree of change in resource transactions represented by the corresponding resource transaction information of the target entity within the target time period. Then, the server 151 can determine the subject type of the target entity based on the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value. For target entities whose subject type belongs to an abnormal subject type (i.e., target entities that conduct abnormal resource transactions), the server 151 can monitor their resource transactions and then issue an alert to a third party (e.g., a regulatory platform) to take predetermined measures against the target entity according to the instructions of the third party (e.g., restricting the target entity's resource transactions within a certain time period, refusing to provide resource transaction services to the target entity, etc.). Of course, the server 151 can directly take these predetermined measures against the target entity.

[0036] In the context of this disclosure, "resources" specifically refers to resources that can be transferred between different entities, particularly resources that can be monetized, such as cash and stocks, but are not limited to physical currency. Resource transaction information refers to information generated by entities conducting resource transaction operations (e.g., resource inflow information, resource outflow information, and resource retention information). For example, when a resource transferor performs a resource inflow operation to transfer resources to a resource transferee, this operation generates resource outflow information if the resource transferor is considered the target entity; and resource inflow information if the resource transferee is considered the target entity. Furthermore, this operation changes the remaining resources (i.e., retained resources) of both the resource transferor and the resource transferee, thus allowing the determination of resource retention information. Additionally, abnormal resource transactions refer to transaction behaviors involved in resource transaction operations or activities that do not comply with relevant legal or ethical standards. For example, abnormal resource transactions could include illegal resource collection operations or illegal large-scale resource outflow operations.

[0037] It should be noted that the descriptions of resource transferors and transferees above are merely illustrative and not restrictive. For example, the resource transferor mentioned above does not necessarily transfer resources to other entities; it may also acquire resources from other entities. Similarly, the resource transferee mentioned above does not necessarily acquire resources from other entities; it may also transfer resources to other entities.

[0038] It should be noted that the term "user" used in this article can refer to any party involved in a resource transaction, including but not limited to human users, online platforms, and even machines.

[0039] In this disclosure, server 151 may be a single server or a cluster of servers on which applications for subject classification and related data can be run. Optionally, server 151 may also run other applications and store other data. For example, server 151 may include multiple virtual hosts for running different applications and providing different services.

[0040] In this disclosure, terminal devices 120 and 130 can be various types of devices, such as mobile phones, tablets, laptops, wearable devices such as smartwatches, in-vehicle devices, etc. Clients can be deployed on terminal devices 120 and 130, which can be used for resource trading operations and optionally provide other services, and can take any of the following forms: locally installed applications, applets accessed via other applications, web applications accessed via a browser, etc. User 110 can view the information presented by the client and perform corresponding interactive operations through the input / output interfaces of terminal devices 120 and 130. Optionally, terminal devices 120 and 130 can be integrated with server 151.

[0041] In this disclosure, the database device 152 can be considered as an electronic filing cabinet, i.e., a place to store electronic files, where users can perform operations such as adding, querying, updating, and deleting data in the files. A "database" is a collection of data stored together in a certain way, which can be shared by multiple objects, has the lowest possible redundancy, and is independent of application programs.

[0042] Furthermore, in this disclosure, network 140 can be a wired network connected via a cable, fiber optic cable, or a wireless network such as 2G, 3G, 4G, 5G, Wi-Fi, Bluetooth, ZigBee, Li-Fi, etc.

[0043] Figure 2 An example flowchart of a subject type determination method 200 according to some embodiments of the present disclosure is illustrated schematically. Exemplarily, the subject type determination method 200 can be... Figure 1 The server 151 shown is used for implementation, but this is not a limitation.

[0044] Specifically, in step 210, resource transaction information of the target entity within a target time period can be obtained. The resource transaction information includes resource transfer-in information, resource transfer-out information, and resource retention information related to the target entity. For example, the resource transfer-in information, resource transfer-out information, and resource retention information can be the transaction flow (i.e., the time series related to the resource transaction information) generated by the target entity conducting corresponding resource transactions within the target time period.

[0045] For example, resource transfer information can take the following forms:

[0046]

[0047] in, This represents the time series corresponding to the resource transfer information. The j-th element in the time series The time interval between any two adjacent elements in the time series can be any value, such as one second, two seconds, one minute, or two minutes. In this case, the target time period and the time series... The length corresponds to, i.e., the time series X i The corresponding time period length is the same as the target time period.

[0048] Similarly, resource transfer information can take the following forms:

[0049]

[0050] in, This represents the time series corresponding to resource transfer information. The j-th element in the time series The time interval between any two adjacent elements in a time series can be any value, such as one second, two seconds, one minute, two minutes, etc. In particular, time series... The time interval between any two adjacent elements in a time series can be correlated with the time series. The time interval between any two adjacent elements in a time series is the same. The corresponding time period length is the same as the target time period.

[0051] Similarly, resource retention information can take the following forms:

[0052]

[0053] in, This represents the time series corresponding to resource retention information. The j-th element in the time series The time interval between any two adjacent elements in a time series can be any value, such as one second, two seconds, one minute, two minutes, etc. In particular, time series... The time interval between any two adjacent elements in a time series can be correlated with the time series. The time interval between any two adjacent elements in a time series is the same. The corresponding time period length is the same as the target time period.

[0054] It should be noted that the acquisition of time series X is affected by factors such as transaction patterns (e.g., transaction frequency, transaction amount) and information acquisition methods. i Time series and time series Subsequently, data cleaning operations may be required (including, but not limited to, outlier removal, missing value imputation, etc.). Unless otherwise explicitly instructed, in this disclosure, time series X i Time series and time series Indicates a time series in which there are no abnormal or missing sequence elements.

[0055] In step 220, resource transfer-in characteristic values ​​corresponding to the resource transfer-in information, resource transfer-out characteristic values ​​corresponding to the resource transfer-out information, and resource retention characteristic values ​​corresponding to the resource retention information can be determined based on the resource transaction information. The resource transfer-in characteristic value, resource transfer-out characteristic value, and resource retention characteristic value are used to represent the degree of change in resource transactions represented by the corresponding resource transaction information of the target entity within the target time period. Specifically, the resource transfer-in characteristic value is used to represent the degree of change in resource transfer-in transactions of the target entity within the target time period; the resource transfer-out characteristic value is used to represent the degree of change in resource transfer-out transactions of the target entity within the target time period; and the resource retention characteristic value is used to represent the degree of change in the retained resources of the target entity within the target time period.

[0056] The applicant discovered that, compared to entities engaging in normal resource transactions, entities engaging in abnormal resource transactions often see these transactions occur within a shorter timeframe. From the perspective of the time series corresponding to the resource transaction information, such abnormal resource transactions cause fluctuations in the corresponding time series within a shorter period. Therefore, analyzing the time series corresponding to resource transaction information allows for the quantification and amplification of the degree of change in resource transactions by a target entity within a target time period using the aforementioned transaction characteristic values, thereby helping to identify whether the target entity is engaging in abnormal resource transactions.

[0057] Time series analysis methods, which measure the complexity and volatility of time series, can be used to determine the transaction characteristic values ​​(i.e., resource inflow characteristic values, resource outflow characteristic values, and resource retention characteristic values) of the target entity based on the resource transaction information. These time series analysis methods include, but are not limited to, calculating the approximate entropy (AE), sample entropy (SE), and permutation entropy (PE) of the time series corresponding to the resource transaction information. Approximate entropy measures the probability of generating new patterns in a signal; more complex time series correspond to larger approximate entropy, while more regular time series correspond to smaller approximate entropy. The calculation of approximate entropy can also be understood as determining the degree of self-similarity of a time series in terms of patterns. For changes in a signal sequence, changes in the approximate entropy value can be used for effective identification. Sample entropy is a statistical measure that does not count self-matches, unlike approximate entropy. Sample entropy is an improvement on the approximate entropy algorithm and is currently a widely used method for calculating entropy characteristic values. Sample entropy is also a measure of time series complexity; the smaller the sample entropy, the smaller the time series complexity and the higher the self-similarity. Permutation entropy is a method for detecting the randomness and dynamic mutation behavior of time series. However, permutation entropy can only detect the complexity and randomness of time series at a single scale. The output time series of complex systems contain feature information at multiple scales. To study the multi-scale complexity changes of time series, the multi-scale permutation entropy (MPE) of the time series can be calculated.

[0058] As described above, the resource transaction information includes resource transfer-in information, resource transfer-out information, and resource retention information related to the target entity; that is, the time series X corresponding to the resource transaction information includes the aforementioned time series. , and .

[0059] In step 230, the subject type of the target entity can be determined based on the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value. That is, step 230 can be used to identify entities engaging in abnormal resource transactions. Any neural network model and machine learning model with classification capabilities can be used to determine the subject type of the target entity based on the transaction characteristic values. Examples of such models include, but are not limited to, fully connected neural networks, decision trees, Naive Bayes, and Support Vector Machines (SVMs).

[0060] It should be noted that various subject types can be set according to the needs of actual applications. For example, the subject type can include a subject type that instructs the subject to conduct abnormal resource transactions (hereinafter referred to as abnormal type for simplicity) and a subject type that instructs the subject to conduct ordinary resource transactions (hereinafter referred to as ordinary type for simplicity). For example, when the subject type includes both abnormal and ordinary types, the process in step 230 can be regarded as a binary classification problem. Taking the SVM model as an example, by constructing a training set with the transaction feature values ​​as sample inputs and the corresponding sample types as sample labels, the SVM model can be trained to meet the preset recognition accuracy. Then, the trained SVM model is used to determine the subject type of the target subject based on the transaction feature values ​​corresponding to the target subject.

[0061] Specifically, a first SVM model corresponding to resource inflow information, a second SVM model corresponding to resource outflow information, and a third SVM model corresponding to resource retention information can be trained separately to determine the subject type of the target subject. The input to the first SVM model can be a resource inflow feature value, the input to the second SVM model can be a resource outflow feature value, and the input to the third SVM model can be a resource retention feature value. For example, these trained SVM models can be used to perform step 230 to determine the subject type of the target subject based on the resource inflow feature value, the resource outflow feature value, and the resource retention feature value. For example, when the output of at least one of the trained first SVM model, second SVM model, and third SVM model indicates that the subject type of the target subject is an anomalous type, the subject type of the target subject can be determined to be an anomalous type.

[0062] Generally, the specific steps of the SVM model (algorithm) are given here:

[0063] Given the original dataset as ,in It is the input feature vector, which in this disclosure is the corresponding transaction feature value (for example, when training the first SVM model, the corresponding transaction feature value is the resource transfer feature value used as the training sample). This refers to the label corresponding to the transaction feature value, which can be of ordinary or abnormal type in this disclosure. i = 1, 2, ..., m represents the number of samples to be classified. An optimal hyperplane needs to be constructed based on the input data source:

[0064]

[0065] Where w and b are preset parameter values, and C is a penalty factor (for example, it can be 1 or other preset values). Let be the relaxed component of the i-th component (representing the degree to which the sample does not satisfy the constraints). By solving the above convex quadratic programming problem, the hyperplane used for classification can be determined. Then, a suitable kernel function is selected ( (e.g., Gaussian kernel function), determined by a classification discriminant function (as shown in the following formula). The category.

[0066]

[0067] pass Figure 2 The subject type determination method 200 shown can determine the resource transfer-in characteristic value corresponding to the resource transfer-in information, the resource transfer-out characteristic value corresponding to the resource transfer-out information, and the resource retention characteristic value corresponding to the resource retention information based on the resource transaction information of the target subject within a target time period. The resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value are respectively used to represent the degree of change in resource transactions represented by the corresponding resource transaction information of the target subject within the target time period. Therefore, the degree of change in resource transactions of the target subject within the target time period can be quantified and amplified through the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value, thereby helping to identify in a timely and accurate manner whether the target subject is engaging in abnormal resource transactions, and thus helping to take appropriate measures against the target subject in a timely manner.

[0068] In some embodiments, resource transfer-in feature values ​​corresponding to resource transfer-in information, resource transfer-out feature values ​​corresponding to resource transfer-out information, and resource retention feature values ​​corresponding to resource retention information are determined based on the resource transaction information. Furthermore, a first multi-scale permutation entropy corresponding to the resource transfer-in information, a second multi-scale permutation entropy corresponding to the resource transfer-out information, and a third multi-scale permutation entropy corresponding to the resource retention information are determined based on the resource transaction information. The first multi-scale permutation entropy is used as the resource transfer-in feature value, the second multi-scale permutation entropy is used as the resource transfer-out feature value, and the third multi-scale permutation entropy is used as the resource retention feature value. For example, the resource transfer-in information, the resource transfer-out information, and the resource retention information can be used as resource processing information, respectively. A multi-scale permutation entropy model with parameters corresponding to the resource processing information is used to determine the first multi-scale permutation entropy, the second multi-scale permutation entropy, and the third multi-scale permutation entropy, respectively, based on the resource processing information. The parameters of the multi-scale permutation entropy model may include signal length, embedding dimension, delay, and scale factor. The values ​​of these parameters in the multi-scale permutation entropy model can be set manually based on experience and application scenarios, or they can be determined using parameter optimization methods in related technologies (e.g., genetic algorithms, particle swarm optimization, ant colony optimization, etc.). Specifically, the embedding dimension can be 3, 4, 5, 6, or 7; the delay can be 1, 2, 3, 4, or 5; and the scale factor can be any integer between 1 and 16 (inclusive). The following is in conjunction with... Figure 3 This section describes in detail the process of determining the multi-scale permutation entropy corresponding to the resource transaction information mentioned above.

[0069] like Figure 3 As shown, in step 310, coarse-graining processing is performed: based on the scale factor, the transaction sequences corresponding to the resource processing information (which are respectively the time series mentioned above) are processed. , and For simplicity, the following text will uniformly represent it as a time series X. Coarse-graining processing is then performed to obtain multiple coarse-grained sequences at different scales corresponding to the transaction sequence. The length of the transaction sequence is the same as the signal length, and the number of sequence elements contained in each of the multiple coarse-grained sequences is the same as the embedding dimension. The specific process is as follows:

[0070] Time series X can have the following form:

[0071]

[0072] in, This represents the j-th element in the time series X corresponding to the resource processing information. The time interval between any two adjacent elements in the time series X can be any value such as one second, two seconds, one minute, or two minutes. In this case, the target time period corresponds to the length of the time series X, that is, the length of the corresponding time period in the time series X is the same as the target time period.

[0073] Multiple coarse-grained sequences at different scales corresponding to time series X can be represented as follows: Generally, the process of calculating the sequence elements in these coarse-grained sequences can be expressed as follows:

[0074]

[0075] Where j=1,2,…,[N / s], [N / s] represents the result of N / s rounded down, and s is the scale factor. Represents coarse-grained sequences at scale s The j-th element. Coarsened sequences at other scales can be obtained through methods similar to... The calculation method is as follows, which will not be elaborated here. For example, coarse-grained sequences at scale 1 can be determined in a similar manner. The j-th element .

[0076] In step 320, phase space reconstruction is performed: multiple coarse-grained sequences are reconstructed respectively. Each coarse-grained sequence in the matrix is ​​reconstructed in phase space to obtain a reconstruction matrix corresponding to each coarse-grained sequence. In any row of the reconstruction matrix, the difference between two adjacent matrix elements and the two sequence indices in the coarse-grained sequence corresponding to the reconstruction matrix is ​​the same as the delay.

[0077] With coarse-grained sequences For example, any row element in the reconstruction matrix obtained by reconstructing its phase space (here, the first row element is taken as the first row element) is considered. (Taking a row element as an example) can be represented as:

[0078]

[0079] in, Represents the first in the above reconstruction matrix row element, m is the embedding dimension. This represents the delay amount.

[0080] The phase space reconstruction results of coarse-grained sequences at other scales can be obtained using a similar approach to the above expression, which will not be elaborated here. For example, coarse-grained sequences at scale 1 can be determined in a similar manner. Any row element in the reconstruction matrix obtained by phase space reconstruction (here, the first row element is referred to as the first row element) row element For example, as shown in the following formula:

[0081]

[0082] In step 330, frequency calculation is performed: for multiple coarse-grained sequences For each coarse-grained sequence, the matrix elements of each row in the reconstruction matrix corresponding to each coarse-grained sequence are arranged in ascending order to obtain the corresponding ascending sequence, and the permutation entropy of the corresponding coarse-grained sequence is determined according to the frequency of the corresponding ascending sequence in the corresponding reconstruction matrix.

[0083] Still using coarse-grained sequences For example, the first element in the reconstruction matrix obtained by reconstructing its phase space is... row element The corresponding ascending sequence obtained after ascending sorting is: Arranging each row of the reconstructed matrix in ascending order will result in multiple ascending sequences. These ascending sequences may be identical; for example, some of these ascending sequences may be identical to... There are 4 identical sequences, and the total number of rows in the reconstruction matrix is ​​. That is, ascending sequence The frequency of occurrence in the corresponding reconstruction matrix can be calculated using the following formula:

[0084]

[0085] Based on the results obtained from the calculation process, the coarse-grained sequence can be determined using the following formula. permutation entropy :

[0086]

[0087] Where V represents the number of distinct ascending sequences among the aforementioned ascending sequences. This represents the frequency of the v-th ascending sequence appearing in the corresponding reconstruction matrix. In fact, in the above calculation of the permutation entropy of this disclosure, frequency is actually used as the probability for calculation.

[0088] Specifically, it can be done using the following formula: Normalization is performed:

[0089]

[0090] Therefore, the above formula can be used to obtain... As a coarse-grained sequence The permutation entropy, to facilitate the differentiation of the permutation entropy of coarse-grained sequences at different scales, can be... permutation entropy Represented as .

[0091] In step 340, multi-scale permutation entropy calculation is performed: based on multiple coarse-grained sequences. The permutation entropy of each coarse-grained sequence in the data is used to determine the multi-scale permutation entropy corresponding to the resource processing information. The calculation of the coarse-grained sequence is shown in the description of step 330 above. permutation entropy The specific process, through a similar process, can yield other coarse-grained sequences. The corresponding permutation entropy Therefore, the multi-scale permutation entropy corresponding to the resource processing information is determined, as shown in the following formula:

[0092]

[0093] By calculating the multi-scale permutation entropy corresponding to the resource processing information, the complex dynamic characteristics of the time series corresponding to the resource processing information can be quantified at multiple different scales. This helps to more accurately analyze the resource transaction operations of the target entity corresponding to the resource processing information within the target time period, and thus more accurately determine the entity type of the target entity.

[0094] Figure 4 An example flowchart illustrating a method for determining the values ​​of parameters of a multi-scale permutation entropy model according to some embodiments of the present disclosure is shown. The parameters are signal length, embedding dimension, delay, and scale factor.

[0095] like Figure 4 As shown, in step 410, the particle swarm is initialized: that is, a particle swarm Z corresponding to the parameters of the multi-scale permutation entropy model is generated.

[0096]

[0097] In this swarm, each particle Z has an initial velocity and an initial position. For example, the j-th particle Z in the swarm Z... j (j=1,2,…,K) has an initial velocity and initial position The initial velocity and initial position of each particle in the particle swarm Z can be random values ​​that follow a preset distribution (e.g., a Gaussian or uniform distribution). The positions of each particle in the particle swarm Z (including the initial position and the positions updated during subsequent iterations) represent different combinations of the aforementioned parameters (signal length, embedding dimension, delay, and scale factor). The number of particles in the particle swarm Z can be determined according to specific requirements, for example, between 20 and 50.

[0098] Based on the initial velocity and initial position of each particle in the particle swarm Z, steps 420-460 are iteratively executed until a preset termination condition is reached to obtain the updated position of each particle, as described in detail below:

[0099] In step 420, the objective function value is calculated: based on the position of each particle in the current iteration, the objective function value of the corresponding particle in the current iteration is calculated. Taking the j-th particle Z in the particle swarm Z as an example... j For example, its position in the current iteration is (Where, variable t indicates the iteration number of the current iteration, i.e., the current iteration is the t-th iteration), then the value of its objective function in the current iteration can be represented by the following formula:

[0100]

[0101] in, Represents the multi-scale permutation entropy in the current iteration. The mean; Represents the multi-scale permutation entropy in the current iteration. The standard deviation; E[] represents the operation of calculating the expectation.

[0102] It should be noted that the above formula actually represents the entropy of multi-scale permutations. The skewness is used as the objective function, but this is merely an example; other forms of objective functions known in the art can be used as needed. Additionally, the multi-scale permutation entropy in the above equation... The calculation process has been described above regarding Figure 3 The description provides a detailed explanation, which will not be repeated here. Additionally, regarding the entropy of multi-scale arrangements... When the skewness is used as the objective function, the corresponding transaction sequence may not necessarily correspond to the resource transaction information generated by the target entity within the target time period. It may also correspond to the resource transaction information generated by an entity with a known entity type within a certain time period.

[0103] In step 430, the optimal position of each particle is determined: the objective function value of each particle in the current iteration is compared with the objective function value of the corresponding particle in the previous iteration to determine the optimal position of each particle. Let's take the j-th particle Z in the particle swarm Z as an example. jFor example, its position in the current iteration is (Where, variable t indicates the iteration number of the current iteration), and its objective function value in the current iteration is... The objective function value With particle Z j The values ​​of the objective function from previous iterations are compared, i.e., a comparison is made. The magnitude of the skewness. When using the skewness of the multi-scale permutation entropy as the objective function, a smaller skewness indicates that the obtained multi-scale permutation entropy is more reliable, that is, it can more accurately represent the fluctuation characteristics of the corresponding time series.

[0104] It should be noted that the individual optimal position of each particle indicates the individual optimal position determined in the current iteration. For example, by comparing... The size of the skewness is used to determine the parameter value of the multi-scale permutation entropy model corresponding to the minimum skewness value, which is then taken as the value of the corresponding particle (here, the j-th particle Z in the particle swarm Z). j The optimal position of an individual particle in the particle swarm Z is determined by the process described above. The optimal positions of other particles in the particle swarm Z in the current iteration can be determined similarly, and will not be repeated here.

[0105] In step 440, the swarm optimal position is determined: based on the individual optimal positions of each particle, the swarm optimal position is determined. For example, based on the individual optimal positions of each particle, the optimal position of the particle with the smallest skewness can be taken as the swarm optimal position. Similar to the meaning of the individual optimal positions of each particle, the swarm optimal position indicates the swarm optimal position determined in the current iteration.

[0106] In step 450, the velocity and position of the particles are updated: based on the individual optimal position of each particle and the group optimal position of the particle swarm, the velocity and position of the corresponding particles are updated. Again, taking the j-th particle Z in the particle swarm Z as an example... j For example, its velocity and position can be updated using the following two formulas:

[0107]

[0108] In the above two formulas, and Let Z represent the particle in the (t+1)th iteration and the tth iteration, respectively. j speed; and Let Z represent the particle in the (t+1)th iteration and the tth iteration, respectively. j Location; and Let Z and Z represent the particle in the t-th iteration, respectively. j The individual optimal position and the group optimal position of the particle swarm Z; For inertial weights, The acceleration constant, These are independent random numbers within the interval (0, 1). The value can be set manually according to requirements, for example... The value can be 0.8, 0.9, 1, etc. The value can be 0.7, 1.5, 2.0, etc. The value can be 0.5, 1.4, 1.9, etc.

[0109] It should be noted that in the description of step 430 above, Represents particle Z j The position of particle Z in the current iteration, and the determination of particle Z in step 430. j The optimal position of the individual in the current iteration, and in step 450, the particle Z in the t-th iteration. j The optimal position of an individual is represented as .

[0110] In step 460, it is determined whether a preset termination condition is met. For example, the preset termination condition may be that the number of iterations has reached a preset number of iterations (e.g., 300 times). Alternatively, the preset termination condition may be that the difference between the objective function value of each particle in the current iteration and the objective function value of the corresponding particle in a previous iteration is less than a first preset threshold (e.g., 0.02). If the preset termination condition is not met, the next iteration is performed, i.e., steps 420-460 are executed again. If the preset termination condition is met, step 470 is executed: based on the updated positions of each particle in the particle swarm Z, the values ​​of the parameters of the multi-scale permutation entropy model are determined. For example, based on the updated positions of each particle in the particle swarm Z, the individual optimal position of each particle can be determined, and then the individual optimal position with the smallest skewness can be taken as the group optimal position of the particle swarm, thereby determining the values ​​of the parameters of the multi-scale permutation entropy model based on the group optimal position of the particle swarm. Alternatively, the values ​​of the parameters of the multi-scale permutation entropy model can be determined based on the individual optimal position of any particle.

[0111] pass Figure 4 The method shown for determining the parameter values ​​of the multi-scale permutation entropy model can make the multi-scale permutation entropy calculated by the multi-scale permutation entropy model more accurately represent the fluctuation characteristics of the corresponding time series. This allows for a more accurate quantification of the complex dynamic characteristics of the time series corresponding to the resource transaction information at multiple different scales, thereby helping to more accurately determine the subject type of the target subject.

[0112] Figure 5An example flowchart illustrating a method for determining transaction characteristic values ​​of a target entity based on resource transaction information according to some embodiments of this disclosure is shown. Figure 5 As shown, the resource transaction information 510 includes resource transfer-in information, resource transfer-out information, and resource retention information related to the target entity.

[0113] like Figure 5 As shown, the transaction feature value 520 includes the resource transfer-in feature value, the resource transfer-out feature value, and the resource retention feature value. For example, a first multi-scale permutation entropy model with parameters corresponding to the resource transfer-in information can be used to determine the first multi-scale permutation entropy as the resource transfer-in feature value based on the resource transfer-in information; a second multi-scale permutation entropy model with parameters corresponding to the resource transfer-out information can be used to determine the second multi-scale permutation entropy as the resource transfer-out feature value based on the resource transfer-out information; and a third multi-scale permutation entropy model with parameters corresponding to the resource retention information can be used to determine the third multi-scale permutation entropy as the resource retention feature value based on the resource retention information.

[0114] It should be noted that the first multi-scale permutation entropy model, the second multi-scale permutation entropy model, and the third multi-scale permutation entropy model can all use the methods described above regarding... Figure 3 The described method determines the first multi-scale permutation entropy, the second multi-scale permutation entropy, and the third multi-scale permutation entropy. The parameters (including signal length, embedding dimension, delay, and scale factor) of the first, second, and third multi-scale permutation entropy models may not be identical. Alternatively, the parameters (including signal length, embedding dimension, delay, and scale factor) of the first, second, and third multi-scale permutation entropy models may be identical. In this case, any one of the first, second, and third multi-scale permutation entropy models can be used to determine the first, second, and third multi-scale permutation entropy.

[0115] pass Figure 5 The method shown for determining the transaction characteristic value of a target entity based on resource transaction information allows for comprehensive consideration of various transaction information related to the target entity's resource transaction operations during the process of determining the entity type. This enables the acquisition of more comprehensive information related to the target entity's resource transaction operations through the corresponding transaction characteristic values, thereby helping to accurately determine the entity type of a target entity engaged in various types of resource transactions.

[0116] Figure 6 The illustration shows an example flowchart of a method for determining the subject type of a target subject based on transaction feature values ​​according to some embodiments of this disclosure. Figure 5 Similarly, in Figure 6 In the method shown, the resource transaction information 610 includes resource transfer-in information, resource transfer-out information, and resource retention information related to the target entity. The transaction feature value 620 obtained by signal processing of the resource transaction information 610 includes resource transfer-in feature value, resource transfer-out feature value, and resource retention feature value. Figure 6 The method shown includes determining the subject type of the target subject based on transaction feature value 620, which includes: determining a first type of the target subject based on resource transfer-in feature value; determining a second type of the target subject based on resource transfer-out feature value; determining a third type of the target subject based on resource retention feature value; and determining the subject type of the target subject based on the first type, the second type, and the third type.

[0117] As described above, the resource inflow characteristic value, the resource outflow characteristic value, and the resource retention characteristic value can all be represented by the multi-scale permutation entropy of the corresponding time series. For example, the resource inflow characteristic value includes a first multi-scale permutation entropy determined based on the resource inflow information; the resource outflow characteristic value includes a second multi-scale permutation entropy determined based on the resource outflow information; and the resource retention characteristic value includes a third multi-scale permutation entropy determined based on the resource retention information.

[0118] Since the resource transfer-in information, resource transfer-out information, and resource retention information are all different, the obtained resource transfer-in characteristic values, resource transfer-out characteristic values, and resource retention characteristic values ​​may also be different. This could lead to the first multi-scale permutation entropy, the second multi-scale permutation entropy, and the third multi-scale permutation entropy being different. In this case, the obtained first type, second type, and third type may not be completely consistent. For example, the obtained first type may indicate that the target entity conducted abnormal resource transactions within the target time period, the obtained second type may indicate that the target entity did not conduct abnormal resource transactions within the target time period, and the obtained third type may also indicate that the target entity did not conduct abnormal resource transactions within the target time period. In this case, the entity type of the target entity can be determined to be an abnormal type based on the first type, the second type, and the third type. That is, as long as at least one of the first type, the second type, and the third type indicates that the target entity conducted abnormal resource transactions within the target time period, the entity type of the target entity can be identified as an abnormal type, thereby enabling more sensitive and accurate identification of the entity type.

[0119] Figure 7 An example flowchart illustrating another method for determining the subject type of a target subject based on transaction feature values ​​according to some embodiments of this disclosure is shown. Figure 7 The basic process shown is as follows Figure 2 The subject type determination method 200 shown is the same, mainly including: performing signal acquisition operations on the activities of the target subject within the target time period to obtain resource transaction information of the target subject within the target time period (the resource transaction information includes resource transfer-in information, resource transfer-out information, and resource retention information related to the target subject); performing signal processing operations on the resource transaction information to determine the transaction characteristic values ​​of the target subject (including resource transfer-in characteristic values, resource transfer-out characteristic values, and resource retention characteristic values) based on the resource transaction information.

[0120] In addition, Figure 7 In the method shown, step 230 of the subject type determination method 200 (i.e., determining the subject type of the target subject based on the resource transfer-in feature value, the resource transfer-out feature value, and the resource retention feature value) includes: taking the resource transfer-in feature value, the resource transfer-out feature value, and the resource retention feature value as resource transaction feature values ​​respectively, and using a trained subject classification model (e.g., an SVM model) corresponding to the resource transaction feature values ​​to determine the first type, the second type, and the third type of the target subject based on the resource transaction feature values ​​respectively.

[0121] For example, the first SVM model, the second SVM model, and the third SVM model described above with respect to step 230 can be used to determine the first type, the second type, and the third type of the target subject, respectively.

[0122] Specifically, the trained subject classification model (i.e., the first SVM model, the second SVM model, and the third SVM model) is trained through the following steps: determining sample transaction feature values ​​corresponding to each sample subject based on multiple sample resource transaction information of multiple sample subjects, each of the multiple sample subjects having a corresponding sample subject type, the multiple sample resource transaction information including one of sample resource transfer-in information, sample resource transfer-out information, and sample resource retention information related to each of the multiple sample subjects, the sample transaction feature value being used to represent the degree of change in resource transactions of the corresponding sample subject within the corresponding sample time period; inputting the sample transaction feature values ​​corresponding to each sample subject into the subject classification model to obtain the predicted subject type corresponding to each sample subject; determining a loss function based on the predicted subject type and the sample subject type corresponding to each sample subject; adjusting the parameters of the subject classification model to minimize the loss function. The loss function can take any functional form known in the art, including but not limited to logarithmic loss function, squared loss function, exponential loss function, absolute value loss function, and cross-entropy loss function.

[0123] For example, historical data on entities engaging in abnormal resource transactions can be obtained from a third party (e.g., a regulatory platform) to form multiple sample entities, and corresponding resource transaction information can be obtained as sample resource transaction information. The multiple sample resource transaction information may include one of the following: sample resource transfer-in information, sample resource transfer-out information, and sample resource retention information related to each of the multiple sample entities, as shown in Table 1:

[0124] Table 1 Sample Resource Transaction Information

[0125] Trading direction No abnormal resource transactions occurred. Abnormal resource transactions occurred Resource transfer Information on sample resource transfer during T1 Information on sample resource transfer during T2 Resource transfer Information on sample resource transfer during T1 Information on sample resource transfer during T2 Resource retention Sample resource retention information during T1 Sample resource retention information during T2

[0126] In Table 1, the time period during which no abnormal resource transactions occurred is denoted as T1, and the time period during which abnormal resource transactions occurred is denoted as T2. T2 can be a time period adjacent to T1. Furthermore, the sample resource transaction information shown in Table 1 (e.g., sample resource transfer-out information during T1) can be represented by a time series of the corresponding transaction flows. The values ​​of T1 and T2 can be determined according to requirements, and this disclosure does not impose any particular restrictions on them. Specifically, the sample resource transaction information used to train the first SVM model may include the sample resource transfer-in information during T1 and the sample resource transfer-in information during T2 shown in Table 1; the sample resource transaction information used to train the second SVM model may include the sample resource transfer-out information during T1 and the sample resource transfer-out information during T2 shown in Table 1; and the sample resource transaction information used to train the third SVM model may include the sample resource retention information during T1 and the sample resource retention information during T2 shown in Table 1.

[0127] As described above regarding step 230, when the subject type includes both abnormal and normal types, the subject type of each of the multiple sample subjects includes both abnormal and normal types. That is, the subject type of the sample subjects corresponding to the sample resource transaction information in Table 1 is either abnormal or normal. For example, the subject types of the sample subjects corresponding to the sample resource transfer-in information, sample resource transfer-out information, and sample resource retention information during T1 are all normal (first type), while the subject types of the sample subjects corresponding to the sample resource transfer-in information, sample resource transfer-out information, and sample resource retention information during T2 are all abnormal (second type). In this case, the degree of change in resource transactions of the sample subjects corresponding to the first type (i.e., normal type) within the corresponding sample time period T1 does not exceed the second preset threshold (e.g., 0.3, 0.5, 0.6, etc.), and the degree of change in resource transactions of the sample subjects corresponding to the second type (i.e., abnormal type) within the corresponding sample time period T2 exceeds the second preset threshold.

[0128] When the sample resource transaction information of multiple sample subjects has the structure shown in Table 1, the values ​​of the parameters (signal length, embedding dimension, delay, and scale factor) of the corresponding multi-scale permutation entropy model can be determined for sample subjects corresponding to different time periods and different subject types, as shown in Table 2:

[0129] Table 2. Parameter combinations for the multi-scale permutation entropy model

[0130] Trading direction No abnormal resource transactions occurred. Abnormal resource transactions occurred Resource transfer The parameter value combination 1 corresponding to T1 The parameter value combination 4 corresponding to T2 Resource transfer The parameter value combination 2 corresponding to T1 The parameter value combination corresponding to T2 is 5. Resource retention The parameter value combination 3 corresponding to T1 The parameter value combination corresponding to T2 is 6.

[0131] In Table 2, the meanings of T1 and T2 are the same as in Table 1. Furthermore, the parameter value combinations shown in Table 2 represent combinations of the parameters (signal length, embedding dimension, delay, and scale factor) of the multi-scale permutation entropy model. Each parameter value combination can be used with the methods described above... Figure 4 The methods described are used to determine this, and will not be elaborated upon here.

[0132] pass Figure 7 The method shown for determining the subject type of a target subject based on transaction feature values ​​utilizes a trained subject classification model to determine the subject type of the target subject based on the transaction feature values. During the training of the subject classification model, the corresponding sample subjects have various types of resource transaction activities (resource transfer in, resource transfer out, resource retention, etc.) and different types of subject types (normal type and abnormal type), thereby enabling the trained subject classification model to accurately determine the subject type of the target subject.

[0133] Figure 8 An example block diagram of a subject type determination device 800 according to some embodiments of the present disclosure is illustrated schematically. Exemplarily, the subject type determination device 800 may be deployed in... Figure 1 On server 151. (For example...) Figure 8 As shown, the subject type determination device 800 includes a resource transaction information acquisition module 810, a transaction feature value determination module 820, and a subject type determination module 830.

[0134] Specifically, the resource transaction information acquisition module 810 can be configured to acquire resource transaction information of the target entity within a target time period. The resource transaction information includes resource transfer-in information, resource transfer-out information, and resource retention information related to the target entity. The transaction feature value determination module 820 can be configured to determine the resource transfer-in feature value corresponding to the resource transfer-in information, the resource transfer-out feature value corresponding to the resource transfer-out information, and the resource retention feature value corresponding to the resource retention information based on the resource transaction information. The resource transfer-in feature value, the resource transfer-out feature value, and the resource retention feature value are used to represent the degree of change in resource transactions represented by the corresponding resource transaction information of the target entity within the target time period. The entity type determination module 830 can be configured to determine the entity type of the target entity based on the resource transfer-in feature value, the resource transfer-out feature value, and the resource retention feature value.

[0135] It should be understood that the subject type determination device 800 can be implemented in software, hardware, or a combination of both. Multiple different modules in the device can be implemented in the same software or hardware structure, or a module can be implemented by multiple different software or hardware structures.

[0136] Furthermore, the subject type determination device 800 can be used to implement the subject type determination method described above, the details of which have been described in detail above and will not be repeated here for the sake of brevity. In addition, these devices can have the same features and advantages as those described in the corresponding methods.

[0137] Figure 9 The illustration depicts an example system including an example computing device 900 representing one or more systems and / or devices that can implement the various technologies described herein. The computing device 900 can be, for example, a server of a service provider, a device associated with a server, a system-on-a-chip, and / or any other suitable computing device or computing system. (Refer to above) Figure 8 The subject type determination device 800 described can take the form of a computing device 900. Alternatively, the subject type determination device 800 can be implemented as a computer program in the form of an application 916.

[0138] like Figure 9 The illustrated example computing device 900 includes a processing system 911 communicatively coupled to each other, one or more computer-readable media 912, and one or more I / O interfaces 913. Although not shown, the computing device 900 may also include a system bus or other data and command transfer system that couples the various components to each other. The system bus may include any or a combination of different bus architectures such as a memory bus or memory controller, a peripheral bus, a universal serial bus, and / or a processor or local bus utilizing any of the various bus architectures. Various other examples, such as control and data lines, are also conceived.

[0139] Processing system 911 represents the functionality of performing one or more operations using hardware. Therefore, processing system 911 is illustrated as including hardware elements 914 that can be configured as processors, function blocks, etc. This may include other logic devices implemented in hardware as application-specific integrated circuits (ASICs) or formed using one or more semiconductors. Hardware element 914 is not limited by the materials in which it is formed or the processing mechanism employed therein. For example, a processor may consist of semiconductors and / or transistors (e.g., integrated circuits (ICs)). In such a context, processor-executable instructions may be electronically executable instructions.

[0140] Computer-readable medium 912 is illustrated as including memory / storage device 915. Memory / storage device 915 represents a memory / storage capacity associated with one or more computer-readable media. Memory / storage device 915 may include volatile media (such as random access memory (RAM)) and / or non-volatile media (such as read-only memory (ROM), flash memory, optical disk, magnetic disk, etc.). Memory / storage device 915 may include fixed media (e.g., RAM, ROM, fixed hard disk drive, etc.) and removable media (e.g., flash memory, removable hard disk drive, optical disk, etc.). Computer-readable medium 912 may be configured in various other ways as further described below.

[0141] One or more I / O interfaces 913 represent the functionality to allow users to input commands and information to the computing device 900 using various input devices and optionally also to present information to the user and / or other components or devices using various output devices. Examples of input devices include keyboards, cursor control devices (e.g., mice), microphones (e.g., for voice input), scanners, touch functionality (e.g., capacitive or other sensors configured to detect physical touch), cameras (e.g., capable of detecting non-touch-related movements as gestures using visible or invisible wavelengths (such as infrared frequencies), etc. Examples of output devices include display devices (e.g., monitors or projectors), speakers, printers, network interface cards, haptic-responsive devices, etc. Therefore, the computing device 900 can be configured to support user interaction in various ways as further described below.

[0142] The computing device 900 also includes an application 916. The application 916 may be, for example, a software instance of the subject type determination device 800, and may be used in combination with other elements in the computing device 900 to implement the techniques described herein.

[0143] This document describes various technologies within the general context of software and hardware components or program modules. Generally, these modules include routines, programs, elements, components, data structures, etc., that perform specific tasks or implement specific abstract data types. As used herein, the terms "module," "function," and "component" generally refer to software, firmware, hardware, or a combination thereof. The technologies described herein are characterized as platform-independent, meaning that these technologies can be implemented on a variety of computing platforms with various processors.

[0144] Implementations of the described modules and technologies may be stored on or transmitted across some form of computer-readable medium. Computer-readable medium may include a variety of media accessible by the computing device 900. By way of example and not limitation, computer-readable medium may include "computer-readable storage medium" and "computer-readable signal medium".

[0145] In contrast to simple signal transmission, carrier waves, or signals themselves, a "computer-readable storage medium" refers to a medium and / or device capable of persistently storing information, and / or a tangible storage device. Therefore, a computer-readable storage medium refers to a non-signal-bearing medium. Computer-readable storage media include hardware such as volatile and non-volatile, removable and non-removable media and / or storage devices implemented using methods or techniques suitable for storing information (such as computer-readable instructions, data structures, program modules, logic elements / circuits, or other data). Examples of computer-readable storage media may include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical storage devices, hard disks, cassette tapes, magnetic tapes, disk storage devices or other magnetic storage devices, or other storage devices, tangible media, or articles of art suitable for storing desired information and accessible by a computer.

[0146] "Computer-readable signal medium" refers to a signal-bearing medium configured to transmit instructions, such as via a network, to computing device 900. A signal medium typically embodies computer-readable instructions, data structures, program modules, or other data in a modulated data signal, such as a carrier wave, data signal, or other transmission mechanism. Signal media also include any information transmission medium. The term "modulated data signal" refers to a signal in which one or more of its characteristics are set or altered to encode information. By way of example and not limitation, communication media include wired media such as wired networks or direct connections, and wireless media such as acoustic, RF, infrared, and other wireless media.

[0147] As previously described, hardware element 914 and computer-readable medium 912 represent instructions, modules, programmable device logic, and / or fixed device logic implemented in hardware, which in some embodiments can be used to implement at least some aspects of the techniques described herein. Hardware elements may include components of integrated circuits or systems-on-a-chip, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), complex programmable logic devices (CPLDs), and other implementations or other hardware devices in silicon. In this context, hardware elements can serve as processing devices for executing program tasks defined by instructions, modules, and / or logic embodied by the hardware element, and as hardware devices for storing instructions for execution, such as the previously described computer-readable storage medium.

[0148] The foregoing combinations can also be used to implement the various techniques and modules described herein. Therefore, software, hardware, or program modules and other program modules can be implemented as one or more instructions and / or logic embodied on some form of computer-readable storage medium and / or by one or more hardware elements 914. The computing device 900 can be configured to implement specific instructions and / or functions corresponding to the software and / or hardware modules. Thus, for example, by using the computer-readable storage medium and / or hardware elements 914 of a processing system, modules can be implemented at least partially in hardware as modules executable as software by the computing device 900. Instructions and / or functions can be executable / operable by one or more articles of art (e.g., one or more computing devices 900 and / or processing systems 911) to implement the techniques, modules, and examples described herein.

[0149] In various implementations, the computing device 900 can be configured in various ways. For example, the computing device 900 can be implemented as a computer-type device, including personal computers, desktop computers, multi-screen computers, laptop computers, netbooks, etc. The computing device 900 can also be implemented as a mobile device, including mobile devices such as mobile phones, portable music players, portable gaming devices, tablet computers, multi-screen computers, etc. The computing device 900 can also be implemented as a television-type device, including devices with or connected to a generally large screen in a leisure viewing environment. These devices include televisions, set-top boxes, game consoles, etc.

[0150] The techniques described herein can be supported by these various configurations of computing device 900, and are not limited to specific examples of the techniques described herein. Functionality can also be implemented, wholly or partially, on the “cloud” 920 using distributed systems, such as through platform 922 as described below.

[0151] Cloud 920 includes and / or represents platform 922 for resource 924. Platform 922 abstracts the underlying functionality of the hardware (e.g., server) and software resources of cloud 920. Resource 924 may include applications and / or data that can be used when performing computer processing on a server located remotely from computing device 900. Resource 924 may also include services provided via the Internet and / or via subscriber networks such as cellular or Wi-Fi networks.

[0152] Platform 922 can abstract resources and functions to connect computing device 900 to other computing devices. Platform 922 can also be used to abstract resource hierarchy to provide a tiered hierarchy of the appropriate level of demand for resources 924 implemented via platform 922. Therefore, in interconnect device embodiments, the implementation of the functions described herein can be distributed throughout system 900. For example, functions can be implemented partly on computing device 900 and partly through platform 922, which abstracts the functions of cloud 920.

[0153] It should be understood that, for clarity, embodiments of this disclosure have been described with reference to different functional units. However, it will be apparent that, without departing from this disclosure, the functionality of each functional unit may be implemented in a single unit, in multiple units, or as part of other functional units. For example, functionality described as being performed by a single unit may be performed by multiple different units. Therefore, references to a particular functional unit are considered merely as references to the appropriate unit used to provide the described functionality, and not as indicating a strict logical or physical structure or organization. Thus, this disclosure may be implemented in a single unit, or may be physically and functionally distributed among different units and circuits.

[0154] It will be understood that although the terms first, second, third, etc., may be used herein to describe various devices, elements, components, or parts, these devices, elements, components, or parts should not be limited by these terms. These terms are used only to distinguish one device, element, component, or part from another device, element, component, or part.

[0155] Although this disclosure has been described in conjunction with some embodiments, it is not intended to be limited to the specific forms set forth herein. Rather, the scope of this disclosure is limited only by the appended claims. Additionally, although individual features may be included in different claims, these may be advantageously combined, and inclusion in different claims does not imply that such a combination of features is not feasible and / or advantageous. The order of features in the claims does not imply that the features must be in any particular order of their operation. Furthermore, in the claims, the word "comprising" does not exclude other elements, and the terms "a" or "an" do not exclude a plurality. Reference numerals in the claims are provided only by way of explicit example and should not be construed as limiting the scope of the claims in any way.

[0156] It should be understood that, for clarity, embodiments of this disclosure have been described with reference to different functional units. However, it will be apparent that, without departing from this disclosure, the functionality of each functional unit may be implemented in a single unit, in multiple units, or as part of other functional units. For example, functionality described as being performed by a single unit may be performed by multiple different units. Therefore, references to a particular functional unit are considered merely as references to the appropriate unit used to provide the described functionality, and not as indicating a strict logical or physical structure or organization. Thus, this disclosure may be implemented in a single unit, or may be physically and functionally distributed among different units and circuits.

[0157] This disclosure provides a computer-readable storage medium storing computer-readable instructions thereon, which, when executed, implement the aforementioned subject type determination method.

[0158] This disclosure provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computing device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computing device to perform the subject type determination method provided in the various optional implementations described above.

[0159] By studying the accompanying drawings, the disclosure, and the appended claims, those skilled in the art can understand and implement variations of the disclosed embodiments in practicing the claimed subject matter. In the claims, the word "comprising" does not exclude other elements or steps, and "a" or "an" does not exclude a plurality. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used advantageously.

Claims

1. A method for determining the type of a subject, characterized in that, The method includes: Obtain resource transaction information of the target entity within a target time period, wherein the resource transaction information includes resource transfer-in information, resource transfer-out information, and resource retention information related to the target entity; Based on the resource transaction information, the following feature values ​​are determined: the resource transfer-in feature value corresponding to the resource transfer-in information, the resource transfer-out feature value corresponding to the resource transfer-out information, and the resource retention feature value corresponding to the resource retention information. These feature values ​​represent the degree of change in resource transactions represented by the corresponding resource transaction information of the target entity within the target time period. The subject type of the target subject is determined based on the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value; The step of determining the resource transfer-in characteristic value corresponding to the resource transfer-in information, the resource transfer-out characteristic value corresponding to the resource transfer-out information, and the resource retention characteristic value corresponding to the resource retention information based on the resource transaction information includes: Based on the resource transaction information, determine the first multi-scale permutation entropy corresponding to the resource transfer-in information, the second multi-scale permutation entropy corresponding to the resource transfer-out information, and the third multi-scale permutation entropy corresponding to the resource retention information; and The first multi-scale permutation entropy is used as the resource-in characteristic value, the second multi-scale permutation entropy is used as the resource-out characteristic value, and the third multi-scale permutation entropy is used as the resource-retention characteristic value.

2. The method according to claim 1, wherein determining the first multi-scale permutation entropy corresponding to the resource transfer-in information, the second multi-scale permutation entropy corresponding to the resource transfer-out information, and the third multi-scale permutation entropy corresponding to the resource retention information based on the resource transaction information comprises: The resource transfer-in information, the resource transfer-out information, and the resource retention information are respectively used as resource processing information. Using a parameterized multi-scale permutation entropy model corresponding to the resource processing information, the first multi-scale permutation entropy, the second multi-scale permutation entropy, and the third multi-scale permutation entropy are determined based on the resource processing information, wherein the parameters of the multi-scale permutation entropy model include signal length, embedding dimension, delay, and scale factor, and include: According to the scale factor, the transaction sequence corresponding to the resource processing information is coarse-grained to obtain multiple coarse-grained sequences at different scales corresponding to the transaction sequence, wherein the length of the transaction sequence is the same as the signal length, and the number of sequence elements contained in each of the multiple coarse-grained sequences is the same as the embedding dimension. Phase space reconstruction is performed on each of the plurality of coarse-grained sequences to obtain a reconstruction matrix corresponding to each coarse-grained sequence. In any row of the reconstruction matrix, the difference between two adjacent matrix elements and the two sequence indices in the coarse-grained sequence corresponding to the reconstruction matrix is ​​the same as the delay amount. For each of the plurality of coarse-grained sequences, the elements of each row of the reconstruction matrix corresponding to each coarse-grained sequence are arranged in ascending order to obtain the corresponding ascending sequence. The permutation entropy of the corresponding coarse-grained sequence is determined based on the frequency of its occurrence in the corresponding reconstruction matrix. Based on the permutation entropy of each of the multiple coarse-grained sequences, the multi-scale permutation entropy corresponding to the resource processing information is determined.

3. The method according to claim 2, wherein the values ​​of the parameters of the multi-scale permutation entropy model are determined by the following steps: Generate a particle swarm corresponding to the parameters of the multi-scale permutation entropy model, wherein each particle in the particle swarm has an initial velocity and an initial position; Based on the initial velocity and initial position of each particle in the particle swarm, the following steps are iteratively executed to reach a preset termination condition, in order to obtain the updated position of each particle: Based on the position of each particle in the current iteration, calculate the value of the objective function for the corresponding particle in the current iteration; Compare the objective function value of each particle in the current iteration with the objective function value of the corresponding particle in the previous iteration to determine the individual optimal position of each particle; Based on the individual optimal position of each particle, the group optimal position of the particle swarm is determined. Based on the individual optimal position of each particle and the group optimal position of the particle swarm, update the velocity and position of the corresponding particles. The values ​​of the parameters of the multi-scale permutation entropy model are determined based on the updated positions of each particle in the particle swarm.

4. The method according to claim 3, wherein the preset termination condition includes one of the following conditions: The number of iterations has reached the preset number of iterations; and The difference between the objective function value of each particle in the current iteration and the objective function value of the corresponding particle in the previous iteration is less than a first preset threshold.

5. The method according to claim 1, wherein determining the subject type of the target subject based on the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value includes: Based on the resource inflow characteristic value, the resource outflow characteristic value, and the resource retention characteristic value, the first type, the second type, and the third type of the target entity are determined respectively; as well as The subject type of the target subject is determined based on the first type, the second type, and the third type.

6. The method according to claim 5, wherein determining the first type, second type, and third type of the target subject based on the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value respectively comprises: The resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value are respectively used as resource transaction characteristic values. Using the trained subject classification model corresponding to the resource transaction characteristic values, the first type, the second type, and the third type of the target subject are determined based on the resource transaction characteristic values. The trained subject classification model is obtained through the following steps: Based on multiple sample resource transaction information of multiple sample subjects, a sample transaction feature value is determined for each sample subject. Each of the multiple sample subjects has a corresponding sample subject type. The multiple sample resource transaction information includes one of the following: sample resource transfer-in information, sample resource transfer-out information, and sample resource retention information related to each of the multiple sample subjects. The sample transaction feature value is used to indicate the degree of change in resource transactions represented by the corresponding resource transaction information of the corresponding sample subject within the corresponding sample time period. Input the sample transaction feature value corresponding to each sample subject into the subject classification model to obtain the predicted subject type corresponding to each sample subject; The loss function is determined based on the prediction subject type and sample subject type corresponding to each sample subject; Adjust the parameters of the subject classification model to minimize the loss function.

7. The method of claim 6, wherein the sample subject type includes a first sample type and a second sample type, and wherein... The degree of change in resource transactions of the sample subject corresponding to the first sample type within the corresponding sample time period does not exceed the second preset threshold, and wherein... The degree of change in resource transactions of the sample subject corresponding to the second sample type within the corresponding sample time period exceeds the second preset threshold.

8. A device for determining the type of a subject, comprising: The resource transaction information acquisition module is configured to acquire resource transaction information of a target entity within a target time period. The resource transaction information includes resource transfer-in information, resource transfer-out information, and resource retention information related to the target entity. The transaction feature value determination module is configured to determine, based on the resource transaction information, resource transfer-in feature values ​​corresponding to the resource transfer-in information, resource transfer-out feature values ​​corresponding to the resource transfer-out information, and resource retention feature values ​​corresponding to the resource retention information. The resource transfer-in feature value, resource transfer-out feature value, and resource retention feature value are respectively used to represent the degree of change in resource transactions represented by the corresponding resource transaction information of the target entity within the target time period; and The subject type determination module is configured to determine the subject type of the target subject based on the resource transfer-in characteristic value, the resource transfer-out characteristic value, and the resource retention characteristic value; The transaction feature value determination module is further configured as follows: Based on the resource transaction information, determine the first multi-scale permutation entropy corresponding to the resource transfer-in information, the second multi-scale permutation entropy corresponding to the resource transfer-out information, and the third multi-scale permutation entropy corresponding to the resource retention information; and The first multi-scale permutation entropy is used as the resource-in characteristic value, the second multi-scale permutation entropy is used as the resource-out characteristic value, and the third multi-scale permutation entropy is used as the resource-retention characteristic value.

9. A computing device, comprising: Memory, which is configured to store computer-executable instructions; A processor configured to perform the method according to any one of claims 1 to 7 when the computer-executable instructions are executed by the processor.

10. A computer-readable storage medium storing computer-executable instructions that, when executed, perform the method according to any one of claims 1 to 7.

11. A computer program product comprising computer-executable instructions, wherein the computer-executable instructions, when executed by a processor, perform the method according to any one of claims 1 to 7.

Images