Target object risk identification method and device

Abstract

The specification relates to the technical field of artificial intelligence, and particularly discloses a target object risk identification method and device, wherein the method comprises: receiving a risk detection request; in response to the risk detection request, obtaining target object data corresponding to a target object identifier; the target object data comprises first type attribute feature data and second type attribute feature data; in the case where there is missing index data and / or erroneous index data in the second type attribute feature data, using a naive Bayes decision algorithm to perform data completion on the missing index data and / or correction on the erroneous index data in the second type attribute feature data, to obtain the second type attribute feature data after completion and / or correction; and performing clustering analysis on the first type attribute feature data and the second type attribute feature data after completion and / or correction, to obtain a risk category corresponding to the target object identifier. The above scheme can improve the accuracy of target object risk identification and adapt to changes in the business environment.

Description

TECHNICAL FIELD

[0001] The present specification relates to the technical field of artificial intelligence, and in particular to a target object risk identification method and device. BACKGROUND

[0002] With the development of Internet technology, e-commerce, social networks, finance, medicine, and scientific and engineering fields have accumulated a large amount of data, showing an exponential growth trend. The vast amount of data contains rich and valuable knowledge, so it is particularly important to extract meaningful and valuable potential information from these complex large-scale data. By learning and analyzing the input data, and using the discovered rules to make reasonable judgments and predictions on the class labels of unknown data, it can be applied to network attack identification, customer churn prediction, earthquake prediction, risk management, medical diagnosis, financial product risk identification, and other application scenarios.

[0003] However, due to the missing or incorrect key indicator data in the input data, it may not be able to adapt to changes in the business environment, resulting in inaccurate risk identification.

[0004] At present, there is no effective solution to the above problem. SUMMARY

[0005] The embodiments of the present specification provide a target object risk identification method and device to solve the problem of inaccurate risk identification in the prior art due to missing or incorrect key indicator data in the input data.

[0006] The embodiments of the present specification provide a target object risk identification method, comprising:

[0007] receiving a risk detection request; the risk detection request carries a target object identifier;

[0008] In response to the risk detection request, obtaining target object data corresponding to the target object identifier; the target object data includes first type attribute feature data and second type attribute feature data; the first type attribute feature data is used to represent the business attribute features associated with the target object; the second type attribute feature data is used to represent the transaction attribute features associated with the target object;

[0009] In the case where the second type attribute feature data has missing indicator data and / or incorrect indicator data, the naive Bayes decision algorithm is used to fill in the missing indicator data and / or correct the incorrect indicator data in the second type attribute feature data, to obtain the second type attribute feature data after filling in and / or correcting;

[0010] performing clustering analysis on the first type of attribute characteristic data and the second type of attribute characteristic data after the data supplementing and / or the error correcting, to obtain a risk category corresponding to the target object identifier.

[0011] In one embodiment, the data supplementing and / or the error correcting of the missing index data in the second type of attribute characteristic data is performed by using a Naive Bayes decision algorithm, including:

[0012] performing preprocessing on the first type of attribute characteristic data and the second type of attribute characteristic data, to obtain preprocessed first type of attribute characteristic data and preprocessed second type of attribute characteristic data;

[0013] detecting the second type of attribute characteristic data to determine whether the second type of attribute characteristic data has missing index data and / or error index data;

[0014] In a case where it is determined that the preprocessed second type of attribute characteristic data has missing index data and / or error index data, the data supplementing and / or the error correcting of the missing index data in the second type of attribute characteristic data is performed by using a Naive Bayes decision algorithm.

[0015] In one embodiment,

[0016] The preprocessed second type of attribute characteristic data includes a plurality of second type of attribute characteristics, and the plurality of second type of attribute characteristics include second type of attribute characteristics with known and correct values and second type of attribute characteristics with unknown or error values.

[0017] Correspondingly, the data supplementing and / or the error correcting of the missing index data in the second type of attribute characteristic data is performed by using a Naive Bayes decision algorithm, including:

[0018] calculating, by using a Naive Bayes decision algorithm, a conditional probability of the second type of attribute characteristics with unknown or error values when the second type of attribute characteristics with known and correct values are taken as a premise;

[0019] based on a value of the second type of attribute characteristics with unknown or error values with the maximum conditional probability, performing the data supplementing and / or the error correcting of the missing index data in the second type of attribute characteristic data.

[0020] In one embodiment, the clustering analysis on the first type of attribute characteristic data and the second type of attribute characteristic data after the data supplementing and / or the error correcting, to obtain a risk category corresponding to the target object identifier, includes:

[0021] perform feature extraction on the first type of attribute feature data and the second type of attribute feature data after the supplement and / or correction to obtain a target feature vector corresponding to the target object identifier;

[0022] calculate distances between the target feature vector and feature vectors corresponding to respective risk categories in the plurality of risk categories;

[0023] determine, as a risk category corresponding to the target object identifier, a risk category corresponding to a cluster center having the smallest distance from the target feature vector.

[0024] In one embodiment, the method further comprises:

[0025] obtain feature vectors corresponding to respective object samples in a large number of object samples; the large number of object samples include object samples of known risk categories;

[0026] perform cluster analysis on the feature vectors corresponding to respective object samples in the large number of object samples to obtain a plurality of cluster centers;

[0027] calculate distances between the feature vectors of the object samples of the known risk categories and respective cluster centers in the plurality of cluster centers to determine risk categories corresponding to respective cluster centers in the plurality of cluster centers.

[0028] In one embodiment, the cluster analysis on the feature vectors corresponding to respective object samples in the large number of object samples to obtain a plurality of cluster centers comprises:

[0029] select, as initial cluster centers, the feature vectors corresponding to respective object samples in a plurality of object samples randomly selected from the large number of object samples to obtain the plurality of cluster centers;

[0030] repeat the following steps until the feature vectors corresponding to the plurality of cluster centers no longer change: calculate distances between the feature vectors corresponding to respective object samples in the large number of object samples except for the object samples corresponding to the plurality of cluster centers and the feature vectors corresponding to respective cluster centers in the plurality of cluster centers to assign to the nearest cluster center to obtain a plurality of clusters; and calculate cluster centers of respective clusters in the plurality of clusters.

[0031] In one embodiment, the cluster analysis on the feature vectors corresponding to respective object samples in the large number of object samples to obtain a plurality of cluster centers comprises:

[0032] calculate average displacements of respective object samples in the large number of object samples;

[0033] perform translation on respective object samples in the large number of object samples;

[0034] Repeat the above steps until the samples converge. Object samples that converge to the same point are identified as object samples of the same cluster, resulting in multiple clusters. Calculate the cluster center corresponding to each of the multiple clusters to obtain multiple cluster centers.

[0035] In one embodiment, the first type of attribute feature data includes at least one of the following data:

[0036] Customer data, usage channel data, processing procedure data, transaction transparency data, and transaction nature data.

[0037] In one embodiment, the second type of attribute feature data includes at least one of the following data:

[0038] Customer number data, high-risk customer number data, number of transactions, and transaction amount data.

[0039] This specification also provides a target object risk identification device, including:

[0040] A receiving module is used to receive risk detection requests; the risk detection request carries a target object identifier.

[0041] The acquisition module is used to acquire target object data corresponding to the target object identifier in response to the risk detection request; the target object data includes a first type of attribute feature data and a second type of attribute feature data; the first type of attribute feature data is used to characterize the business attribute features associated with the target object; the second type of attribute feature data is used to characterize the transaction attribute features associated with the target object;

[0042] The data completion and correction module is used to complete the missing indicator data and / or correct the erroneous indicator data in the second type of attribute feature data when there are missing indicator data and / or erroneous indicator data in the second type of attribute feature data, so as to obtain the completed and / or corrected second type of attribute feature data.

[0043] The clustering analysis module is used to perform clustering analysis on the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data to obtain the risk category corresponding to the target object identifier.

[0044] This specification also provides a computer device, including a processor and a memory for storing processor-executable instructions, wherein the processor executes the instructions to implement the steps of the target object risk identification method described in any of the above embodiments.

[0045] This specification also provides a computer-readable storage medium storing computer instructions that, when executed, implement the steps of the target object risk identification method described in any of the above embodiments.

[0046] This specification provides a method for identifying the risk of a target object. A server can receive a risk detection request from a client. In response to the request, the server can obtain target object data corresponding to a target object identifier. This data can be first-class attribute feature data and second-class attribute feature data of the target object. The server can use a Naive Bayes decision algorithm to complete missing or erroneous indicator data in the second-class attribute feature data, obtaining completed and / or corrected second-class attribute feature data. Then, the server can perform cluster analysis on the first-class attribute feature data and the completed and / or corrected second-class attribute feature data to obtain the risk category corresponding to the target object identifier. In this scheme, by assessing the risk of the target object from both first-class and second-class attribute feature data, the features contained in the target object data are more comprehensive, thus improving the accuracy of risk identification. Furthermore, by using the Naive Bayes decision algorithm to complete or correct the second-type attribute feature data in the target object data, the integrity and accuracy of the data in risk identification can be guaranteed. Moreover, the Naive Bayes decision algorithm is more efficient and faster than other algorithms. After obtaining complete and accurate object data, cluster analysis can be performed on the object data to obtain the risk category corresponding to the target object, which can improve the accuracy of target object risk identification, improve transaction security, protect user rights, and improve user experience. Attached Figure Description

[0047] The accompanying drawings, which are included to provide a further understanding of this specification and form part of it, do not constitute a limitation thereof. In the drawings:

[0048] Figure 1 A schematic diagram illustrating an application scenario of the target object risk identification method in one embodiment of this specification is shown;

[0049] Figure 2 A flowchart of a target object risk identification method according to an embodiment of this specification is shown;

[0050] Figure 3 A schematic diagram of the structure of an apparatus for implementing a target object risk identification method according to an embodiment of this specification is shown;

[0051] Figure 4 A flowchart of a target object risk identification method according to an embodiment of this specification is shown;

[0052] Figure 5 A schematic diagram of a target object risk identification device according to one embodiment of this specification is shown;

[0053] Figure 6 A schematic diagram of a computer device according to one embodiment of this specification is shown. Detailed Implementation

[0054] The principles and spirit of this specification will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are given merely to enable those skilled in the art to better understand and implement this specification, and are not intended to limit the scope of this specification in any way. Rather, these embodiments are provided to make this disclosure more thorough and complete, and to fully convey the scope of this disclosure to those skilled in the art.

[0055] Those skilled in the art will recognize that the embodiments described in this specification can be implemented as a system, apparatus, method, or computer program product. Therefore, the disclosure of this specification can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0056] This specification provides a method for identifying the risk of a target object through its embodiments. Figure 1 This diagram illustrates an application scenario of the target object risk identification method according to one embodiment of this specification. In one scenario example, such as... Figure 1 As shown, the method in this embodiment can be applied to a server. The server can receive a risk detection request sent by the client. The risk detection request may carry a target object identifier. In response to the risk detection request, the server can obtain the target object data corresponding to the target object identifier. The target object data may be the first type of attribute feature data and the second type of attribute feature data of the target object.

[0057] The first type of attribute feature data is used to characterize the business attribute features associated with the target object. In one embodiment, the first type of attribute feature data may include at least one of the following: customer object data, usage channel data, processing procedure data, transaction transparency data, and transaction nature data. The second type of attribute feature data is used to characterize the transaction attribute features associated with the target object. In one embodiment, the second type of attribute feature data may include at least one of the following: number of customers, number of high-risk customers, number of transactions, and transaction amount data.

[0058] The server can use the Naive Bayes decision algorithm to impute missing or erroneous indicator data in the second-type attribute feature data, obtaining the imputed and / or corrected second-type attribute feature data. Then, the server can perform cluster analysis on the first-type attribute feature data and the imputed and / or corrected second-type attribute feature data to obtain the risk category corresponding to the target object identifier.

[0059] The aforementioned server can be a single server, a server cluster, or a cloud server, etc., and the specific composition is not limited in this application. The aforementioned client can be a desktop computer, laptop, mobile terminal, PDA, etc., as long as it is a device that can display content to users or business personnel and receive operation instructions. The presentation of the client is also not limited in this application.

[0060] Figure 2 A flowchart of a target object risk identification method according to one embodiment of this specification is shown. Although this specification provides method operation steps or apparatus structures as shown in the following embodiments or figures, more or fewer operation steps or module units may be included in the method or apparatus based on conventional or non-inventive effort. In steps or structures where there is no logically necessary causal relationship, the execution order of these steps or the module structure of the apparatus is not limited to the execution order or module structure described in the embodiments and figures of this specification. When the method or module structure is applied in actual devices or terminal products, it can be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment, or even a distributed processing environment) according to the method or module structure shown in the embodiments or figures.

[0061] Specifically, such as Figure 2 As shown, one embodiment of this specification provides a target object risk identification method that may include the following steps.

[0062] Step S201: Receive a risk detection request; the risk detection request carries a target object identifier.

[0063] Step S202: In response to the risk detection request, obtain the target object data corresponding to the target object identifier; the target object data includes a first type of attribute feature data and a second type of attribute feature data; the first type of attribute feature data is used to characterize the business attribute features associated with the target object; the second type of attribute feature data is used to characterize the transaction attribute features associated with the target object.

[0064] The target object risk identification method in this embodiment can be applied to a server. The server can receive risk detection requests. The risk detection request may carry a target object identifier. The target object identifier may be the identification information of the target object to be classified into risk categories. The target object may be a financial product such as a wealth management product.

[0065] In response to a risk detection request, the server can retrieve target object data corresponding to the target object identifier. In one embodiment, the server can send a retrieval request to the database, which may carry the target object identifier. The server can receive the target object data returned by the database. The target object data may include first-type attribute feature data and second-type attribute feature data of the target object.

[0066] The first type of attribute feature data is used to characterize the business attribute features associated with the target object, and can qualitatively assess the risk of the target object.

[0067] The second type of attribute feature data is used to characterize the transaction attribute features associated with the target object, and can quantitatively characterize the risk of the target object.

[0068] In a financial context, the target object can be a financial product. Accordingly, in some embodiments of this specification, the first type of attribute characteristic data may include at least one of the following: customer target data, usage channel data, processing procedure data, transaction transparency data, and transaction nature data. Customer target data may refer to whether the target object is a product developed for a specific customer group. Usage channel data may refer to whether the target object supports face-to-face transactions. The processing procedure may characterize the complexity of the target object's first-time application and the complexity of subsequent applications. Transaction transparency data may characterize the transparency of counterparty information and the transparency of transaction-related parties. Transaction nature data may characterize whether the target object's transactions support interbank transactions, cross-border transactions, and foreign currency transactions.

[0069] In some embodiments of this specification, the second type of attribute feature data may include at least one of the following: customer quantity data, high-risk customer quantity data, transaction count, and transaction amount data. Specifically, the customer quantity data may include the number of customers corresponding to each time period of the target object across multiple time periods. The high-risk customer quantity data may include the number of high-risk customers corresponding to each time period of the target object across multiple time periods. The transaction count data may include the number of transactions corresponding to each time period of the target object across multiple time periods. The transaction amount data may include the transaction amount corresponding to each time period of the target object across multiple time periods.

[0070] Step S203: In the case that there are missing and / or incorrect indicator data in the second type of attribute feature data, the Naive Bayes decision algorithm is used to complete the missing indicator data and / or correct the incorrect indicator data in the second type of attribute feature data, so as to obtain the completed and / or corrected second type of attribute feature data.

[0071] The server can use the Naive Bayes decision algorithm to complete missing indicator data in the second type of attribute feature data and / or correct erroneous indicator data in the second type of attribute feature data, so as to obtain the completed and / or corrected second type of attribute feature data.

[0072] In some embodiments of this specification, the Naive Bayes decision algorithm is used to complete missing indicator data and / or correct erroneous indicator data in the second type of attribute feature data. This includes: preprocessing the first type of attribute feature data and the second type of attribute feature data to obtain preprocessed first type of attribute feature data and preprocessed second type of attribute feature data; detecting the second type of attribute feature data to determine whether there is missing indicator data and / or erroneous indicator data; and, if it is determined that there is missing indicator data and / or erroneous indicator data in the preprocessed second type of attribute feature data, using the Naive Bayes decision algorithm to complete the missing indicator data and / or correct the erroneous indicator data in the second type of attribute feature data.

[0073] The server can preprocess the first-type and second-type attribute feature data to obtain preprocessed first-type and second-type attribute feature data. Specifically, the server can perform data cleaning on the first-type and second-type attribute feature data, such as currency conversion, standardizing quantity units, and transaction amount units. The server can filter the first-type and second-type attribute feature data by judging whether the indicator data is empty to filter complete data and missing indicator data. The server can also determine whether there is erroneous second-type attribute feature data based on second-type attribute feature data from multiple time periods. After identifying missing and / or erroneous second-type attribute feature data, the Naive Bayes decision algorithm can be used to complete the missing indicator data and / or correct the erroneous indicator data in the second-type attribute feature data.

[0074] In some embodiments of this specification, the preprocessed second-type attribute feature data includes multiple second-type attribute features, including second-type attribute features with known and correct values ​​and second-type attribute features with unknown or incorrect values. Correspondingly, a Naive Bayes decision algorithm is used to complete missing indicator data and / or correct incorrect indicator data in the second-type attribute feature data. This includes: using the Naive Bayes decision algorithm to calculate the conditional probability of the unknown or incorrect second-type attribute features taking different values, given that the second-type attribute features with known and correct values ​​are present; and based on the value of the unknown or incorrect second-type attribute feature with the highest conditional probability, completing missing indicator data and / or correcting incorrect indicator data in the second-type attribute feature data.

[0075] In this embodiment, the preprocessed second-type attribute feature data includes multiple second-type attribute features. These multiple second-type attribute features include those with known and correct values, and those with unknown or incorrect values. The Naive Bayes algorithm is an extension of the Bayesian algorithm, making several assumptions to improve its efficiency and speed. Naive Bayes uses the principle of conditional probability and assumes that the feature conditions are independent. It first learns the joint probability distribution from input to output using a given training set, assuming the independence of feature values. Input X yields the output Y that maximizes the posterior probability. The Naive Bayes decision algorithm can be used to calculate the conditional probabilities of the unknown or incorrect second-type attribute features taking different values, assuming the known and correct values ​​of the second-type attribute features. Then, based on the value of the unknown or incorrect second-type attribute feature with the highest conditional probability, data completion can be performed on missing indicator data and / or incorrect indicator data can be corrected in the second-type attribute feature data.

[0076] Step S204: Perform cluster analysis on the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data to obtain the risk category corresponding to the target object identifier.

[0077] After supplementing and / or correcting the second type of attribute feature data, cluster analysis can be performed on the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data to obtain the risk category corresponding to the target object identifier.

[0078] In some embodiments of this specification, cluster analysis is performed on the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data to obtain the risk category corresponding to the target object identifier. This includes: extracting features from the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data to obtain the target feature vector corresponding to the target object identifier; calculating the distance between the target feature vector and the feature vectors of the cluster centers corresponding to each of the multiple risk categories; and determining the risk category corresponding to the cluster center with the smallest distance to the target feature vector as the risk category corresponding to the target object identifier.

[0079] In this embodiment, feature extraction can be performed on the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data of the target object to obtain the target feature vector corresponding to the target object identifier. In one embodiment, the multiple cluster centers can be cluster center 1, cluster center 2, cluster center 3, cluster center 4, and cluster center 5. The risk category corresponding to cluster center 1 is high-risk, the risk category corresponding to cluster center 2 is medium-high risk, the risk category corresponding to cluster center 3 is medium risk, the risk category corresponding to cluster center 4 is medium-low risk, and the risk category corresponding to cluster center 5 is low risk. The distance between the target feature vector and the feature vectors corresponding to each of the five cluster centers can be calculated, and the risk category corresponding to the cluster center with the smallest distance to the target feature vector is determined as the risk category corresponding to the target object identifier. For example, if the distance between the target feature vector and the feature vector corresponding to cluster center 3 is the smallest, then the risk category of the target object can be determined as medium risk.

[0080] In the above embodiments, by assessing the risk of the target object from both the first type of attribute feature data and the second type of attribute feature data, the features contained in the target object data are made more comprehensive, thus improving the accuracy of risk identification. Furthermore, by using the Naive Bayes decision algorithm to complete or correct the second type of attribute feature data in the target object data, the completeness and correctness of the data during risk identification can be guaranteed. Moreover, the Naive Bayes decision algorithm is more efficient and faster than other algorithms. After obtaining complete and correct object data, cluster analysis can be performed on the object data to obtain the risk category corresponding to the target object, which can improve the accuracy of target object risk identification, enhance transaction security, protect user rights, and improve user experience.

[0081] In some embodiments of this specification, the method may further include: obtaining feature vectors corresponding to each object sample in a large number of object samples; the large number of object samples includes object samples with known risk categories; performing cluster analysis on the feature vectors corresponding to each object sample in the large number of object samples to obtain multiple cluster centers; calculating the distance between the feature vectors of the object samples with known risk categories and each cluster center in the multiple cluster centers to determine the risk category corresponding to each cluster center in the multiple cluster centers.

[0082] Specifically, the feature vectors corresponding to the cluster centers of each cluster in multiple clusters can be predetermined. The feature vectors corresponding to each object sample in a large number of object samples can be obtained. Cluster analysis can be performed on the feature vectors corresponding to each object sample in the large number of object samples to obtain multiple cluster centers. The risk types corresponding to multiple object samples in the large number of object samples are known. The distance between the feature vector of the object sample with a known risk category and each cluster center in the multiple cluster centers can be calculated. For example, object sample A corresponds to a high-risk category, object sample B corresponds to a medium-high-risk category, object sample C corresponds to a medium-risk category, object sample D corresponds to a medium-low-risk category, and object sample E corresponds to a low-risk category. The multiple cluster centers can be cluster center 1, cluster center 2, cluster center 3, cluster center 4, and cluster center 5. After calculating the distances, it is found that sample A is closest to cluster center 1, sample B is closest to cluster center 2, sample C is closest to cluster center 3, sample D is closest to cluster center 4, and sample E is closest to cluster center 5. Therefore, cluster center 1 corresponds to a high-risk category, cluster center 2 to a medium-high-risk category, cluster center 3 to a medium-risk category, cluster center 4 to a medium-low-risk category, and cluster center 5 to a low-risk category. This method can be used to determine the risk categories corresponding to multiple cluster centers.

[0083] In some embodiments of this specification, cluster analysis is performed on the feature vectors corresponding to each object sample in the large number of object samples to obtain multiple cluster centers. This may include: randomly selecting the feature vectors corresponding to each object sample in the large number of object samples as initial cluster centers to obtain multiple cluster centers; repeating the following steps until the feature vectors corresponding to the multiple cluster centers no longer change; calculating the distance between the feature vectors corresponding to other object samples in the large number of object samples besides the object samples corresponding to the multiple cluster centers and the feature vectors corresponding to each cluster center in the multiple cluster centers, and assigning them to the nearest cluster center to obtain multiple clusters; and calculating the cluster center of each cluster in the multiple clusters.

[0084] In this embodiment, the K-Means algorithm can be used to perform cluster analysis on the feature vectors corresponding to each object sample in a large number of object samples to obtain multiple cluster centers. Specifically, the feature vectors corresponding to each object sample in a large number of object samples can be randomly selected as initial cluster centers to obtain multiple cluster centers. The distance between the feature vectors corresponding to other object samples in the large number of object samples (excluding the object samples corresponding to the multiple cluster centers) and the feature vectors corresponding to each of the multiple cluster centers is calculated, and the samples are assigned to the nearest cluster center to obtain multiple clusters; the cluster center of each cluster is calculated. Each time a sample is assigned, the cluster center of the cluster is recalculated based on the existing objects in the cluster. This process is repeated until no object is reassigned to a different cluster and no cluster center changes.

[0085] In some embodiments of this specification, cluster analysis is performed on the feature vectors corresponding to each object sample in the large number of object samples to obtain multiple cluster centers. This may include: calculating the average displacement of each object sample in the large number of object samples; translating each object sample in the large number of object samples; repeating the above steps until the samples converge, and determining the object samples that converge to the same point as object samples of the same cluster to obtain multiple clusters; calculating the cluster center corresponding to each cluster in the multiple clusters to obtain multiple cluster centers.

[0086] In this embodiment, the Mean Shift algorithm can be used to perform cluster analysis on the feature vectors corresponding to each object sample in a large number of object samples, resulting in multiple cluster centers. The average displacement of each object sample in the large number of object samples can be calculated; the object samples in the large number of object samples can be translated. The above steps are repeated until the samples converge. Object samples that converge to the same point are determined as object samples of the same cluster, resulting in multiple clusters. The cluster centers corresponding to each of the multiple clusters are calculated, resulting in multiple cluster centers.

[0087] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. For details, please refer to the foregoing descriptions of the relevant processing embodiments; they will not be repeated here.

[0088] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0089] The above method will be described below with reference to a specific embodiment. However, it is worth noting that this specific embodiment is only for better illustration of this specification and does not constitute an improper limitation of this specification.

[0090] This specific embodiment provides a method for classifying product line levels. Please refer to... Figure 3 The diagram shows a structural schematic of an apparatus for implementing the method in this specific embodiment. Figure 3 As shown, it may include a qualitative assessment data acquisition module 101, a quantitative indicator data acquisition module 102, a basic data processing module 103, a product classification module 104, and a product classification data output module 105.

[0091] The qualitative assessment data acquisition module 101 can generate a product qualitative data table based on the product's customer base, usage channels, processing procedures, transaction transparency, and transaction nature, as shown in Table 1.

[0092] Table 1

[0093]

[0094] The quantitative indicator data acquisition module 102 generates a product quantitative data table based on four indicators in the time dimension: number of customers, number of high-risk customers, number of transactions, and transaction amount, as shown in Table 2.

[0095] Table 2

[0096] Time period 1 Time period 2 Time period 3 Number of customers a1 a2 a3 Number of high risk customers b1 b2 b3 Number of transactions c1 c2 c3 Transaction amount d1 d2 d3

[0097] The basic data processing module 103 can perform data cleaning based on the raw data acquired by the qualitative evaluation data acquisition module 101 and the quantitative indicator data acquisition module 102, clarifying the quantity measurement unit (units, tens of thousands), transaction amount unit, and currency conversion, etc. Then, through the Naive Bayes decision algorithm, it predicts missing product key indicator data, fills in missing data, and corrects erroneous data, ensuring the integrity and accuracy of the data when it is accessed by the product classification module, such as... Figure 4 The flowchart shown is a product risk level classification method in this embodiment.

[0098] Logistic regression is a commonly used machine learning algorithm, generally used to estimate the probability of something, and can also be used for classification and prediction. Regression is essentially estimating unknown parameters of a known formula; it's a linear regression normalized by the regression equation. However, for data with complex features, it's prone to underfitting, resulting in mediocre performance. It's often used for simpler natural language processing tasks, typically when features are well-defined and few in number. Compared to logistic regression, Naive Bayes offers more stable classification efficiency and higher speed for training and querying large datasets.

[0099] The Bayesian algorithm is developed based on the probabilistic model of Thomas Bayes. It's a probabilistic model that calculates the probability distribution of unknowns after acquiring current observation data, and it can update existing results based on new observation data. The Naive Bayes algorithm is an extension of the Bayesian algorithm, making some assumptions to improve its efficiency and speed.

[0100] Naive Bayes is a method based on the principle of conditional probability and assuming that the features are independent of each other. It first learns the joint probability distribution from input to output using a given training set, with the premise that the features are independent. It then calculates the output Y that maximizes the posterior probability from the input X.

[0101] Derivation of the general form of Bayes' theorem:

[0102]

[0103] Where P(A,B) is the joint probability, P(A|B) is the marginal probability, and P(A) is the conditional probability.

[0104] Applying Bayes' criterion to conditional probability yields:

[0105] If P(c1|x,y)>P(c2|x,y), then it belongs to c1;

[0106] If P(c2|x,y)>P(c1|x,y), then it belongs to category c2.

[0107] like Figure 4 As shown, the method in this specific embodiment may include the following.

[0108] (1) Data input: Qualitative evaluation data acquisition module 101 and quantitative indicator data acquisition module 102 serve as the raw data input to module 103.

[0109] (2) Data cleaning, currency conversion, and standardization of quantity units (units, ten thousand units), transaction amount units, etc.

[0110] (3) Data filtering: filter complete data and missing indicator data by judging whether the indicator data is empty.

[0111] (4) Data prediction: Use complete counterparty information data as training sample data to construct a training sample data set.

[0112]

[0113] According to the above formula, the probability of event B given event A can be calculated using the probability of event A, the probability of event B, and the probability of event A given event B. The three probabilities on the right side of the equation are prior probabilities, which can be obtained from the training set. Therefore, the problem is transformed into finding P(A|B) and P(B). In practical applications, a feature can have an infinite number of values. If a joint probability distribution is used, the prior probability needs to be calculated for every combination of every value for every feature, which is practically infeasible.

[0114] To address this problem, we can assume conditional independence for the conditional probability distribution. When events are independent, P(A,B) = P(A)P(B), and the original conditional probability can be transformed into the product of the conditional probabilities of multiple independent events:

[0115]

[0116] Under this assumption, the posterior probability is calculated using Bayes' theorem:

[0117]

[0118] Since x is the feature vector of the new input instance, the denominator remains unchanged for any value of Y. The class with the highest posterior probability is output as the class of x. When comparing sizes, identical denominators can be ignored. Therefore, the Naive Bayes formula can be transformed into:

[0119]

[0120] Where P(Y=Ck) is the prior probability, and P(X(j)=x(j)|Y=Ck) is the conditional probability, which can be derived from the training set. The above formula represents calculating the probability that an instance belongs to a certain class given a certain feature vector, and the class corresponding to the highest probability is considered the class of that point. The problem can be transformed into calculating the prior probability and conditional probability. The method for calculating the prior probability is as follows:

[0121]

[0122] Where N is the total number of instances in the training set, and I is an indicator function that takes the value 1 if the condition in parentheses is true, and 0 otherwise.

[0123] Based on the method derived above, the indicator data corresponding to the highest probability value is taken as the predicted value, and the missing indicator data is supplemented.

[0124] (5) Data completion: Complete the missing indicator data.

[0125] The product classification module 104 obtains complete and valid product data to be classified based on the basic data processing module 103, and uses the K-Means clustering classification algorithm to classify products into five categories: low risk, low-medium risk, medium risk, medium-high risk, and high risk.

[0126] The K-Means algorithm is an iterative clustering algorithm. It randomly selects K objects as initial cluster centers, then calculates the distance between each object and each seed cluster center, assigning each object to the nearest cluster center. Each time a sample is assigned, the cluster centers are recalculated based on the existing objects in the cluster. This process is repeated until no object is reassigned to a different cluster and no cluster centers change.

[0127] Algorithm steps:

[0128] The input is a sample set D = {x1, x2, ..., x} m}, where k is the cluster tree for clustering, and N is the maximum number of iterations.

[0129] The output is a cluster partition C = {C1, C2, ..., C} k}

[0130] (1) Randomly select k samples from dataset D as the initial k centroid vectors {u1, u2, ..., u...} k}

[0131] (2) For n = 1, 2, ... N.

[0132] a) Initialize the cluster partition C to... t = 1, 2, ..., k.

[0133] b) For i = 1, 2, ..., m, calculate sample x. i and each centroid vector u j Distance (j=1,2,...k) x i The smallest one is d. ij The corresponding category λ i Update now

[0134] c) For j = 1, 2...k, for C jRecalculate the new centroid for all sample points.

[0135] d) If none of the k centroid vectors have changed, proceed to step (3).

[0136] (3) Output cluster partitioning C = {C1, C2, ..., C} k}

[0137] The method in this embodiment overcomes the problems of missing key product indicator data and inaccurate product classification levels in existing methods, improving the completeness and accuracy of financial product data and enhancing the targeting and effectiveness of anti-money laundering work. Through qualitative assessment data and quantitative indicator data collection modules, various data of financial products are preserved more effectively and permanently, eliminating reliance on manual storage, thus reducing human resource investment and improving work efficiency. Compared to rule-based product classification methods, this embodiment relies less on pre-defined rules and can better adapt to changes in the business environment. The product money laundering classification levels are more accurate and error-free, guiding financial institutions to more rationally apply the "risk-based" concept and adopt different risk prevention measures, effectively preventing and mitigating money laundering risks for certain product categories.

[0138] Based on the same inventive concept, this specification also provides a target object risk identification device in its embodiments, as described in the following embodiments. Since the principle by which the target object risk identification device solves the problem is similar to that of the target object risk identification method, the implementation of the target object risk identification device can refer to the implementation of the target object risk identification method, and repeated details will not be elaborated further. As used below, the terms "unit" or "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated. Figure 5 This is a structural block diagram of a target object risk identification device according to an embodiment of this specification, such as... Figure 5 As shown, it includes: a receiving module 501, an acquisition module 502, a completion and correction module 503, and a clustering analysis module 504. The structure is described below.

[0139] The receiving module 501 is used to receive risk detection requests; the risk detection requests carry the target object identifier.

[0140] The acquisition module 502 is used to acquire target object data corresponding to the target object identifier in response to the risk detection request; the target object data includes a first type of attribute feature data and a second type of attribute feature data; the first type of attribute feature data is used to characterize the business attribute features associated with the target object; the second type of attribute feature data is used to characterize the transaction attribute features associated with the target object.

[0141] The completion and correction module 503 is used to complete the missing indicator data and / or correct the erroneous indicator data in the second type of attribute feature data by using the Naive Bayes decision algorithm when there are missing indicator data and / or erroneous indicator data in the second type of attribute feature data, so as to obtain the completed and / or corrected second type of attribute feature data.

[0142] The clustering analysis module 504 is used to perform clustering analysis on the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data to obtain the risk category corresponding to the target object identifier.

[0143] In some embodiments of this specification, the completion and correction module is specifically used for: preprocessing the first type of attribute feature data and the second type of attribute feature data to obtain preprocessed first type of attribute feature data and preprocessed second type of attribute feature data; detecting the second type of attribute feature data to determine whether there are missing indicator data and / or erroneous indicator data in the second type of attribute feature data; and, if it is determined that there are missing indicator data and / or erroneous indicator data in the preprocessed second type of attribute feature data, using a Naive Bayes decision algorithm to complete the missing indicator data and / or correct the erroneous indicator data in the second type of attribute feature data.

[0144] In some embodiments of this specification, the preprocessed second-type attribute feature data includes multiple second-type attribute features, including second-type attribute features with known and correct values ​​and second-type attribute features with unknown or incorrect values. Accordingly, the completion and correction module is specifically used to: calculate the conditional probability of the unknown or incorrect second-type attribute features taking different values ​​under the premise that the second-type attribute features with known and correct values ​​are known; and based on the value of the unknown or incorrect second-type attribute feature with the highest conditional probability, perform data completion for missing indicator data and / or correction for incorrect indicator data in the second-type attribute feature data.

[0145] In some embodiments of this specification, cluster analysis is performed on the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data to obtain the risk category corresponding to the target object identifier. This includes: extracting features from the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data to obtain the target feature vector corresponding to the target object identifier; calculating the distance between the target feature vector and the feature vectors of the cluster centers corresponding to each of the multiple risk categories; and determining the risk category corresponding to the cluster center with the smallest distance to the target feature vector as the risk category corresponding to the target object identifier.

[0146] In some embodiments of this specification, the apparatus further includes a clustering module, which is specifically used for: obtaining feature vectors corresponding to each object sample in a large number of object samples; the large number of object samples includes object samples with known risk categories; performing cluster analysis on the feature vectors corresponding to each object sample in the large number of object samples to obtain multiple cluster centers; calculating the distance between the feature vectors of the object samples with known risk categories and each cluster center in the multiple cluster centers to determine the risk category corresponding to each cluster center in the multiple cluster centers.

[0147] In some embodiments of this specification, cluster analysis is performed on the feature vectors corresponding to each object sample in the large number of object samples to obtain multiple cluster centers. This includes: randomly selecting the feature vectors corresponding to each object sample in the large number of object samples as initial cluster centers to obtain multiple cluster centers; repeating the following steps until the feature vectors corresponding to the multiple cluster centers no longer change; calculating the distance between the feature vectors corresponding to other object samples in the large number of object samples besides the object samples corresponding to the multiple cluster centers and the feature vectors corresponding to each cluster center in the multiple cluster centers, and assigning them to the nearest cluster center to obtain multiple clusters; and calculating the cluster center of each cluster in the multiple clusters.

[0148] In some embodiments of this specification, cluster analysis is performed on the feature vectors corresponding to each object sample in the large number of object samples to obtain multiple cluster centers. This includes: calculating the average displacement of each object sample in the large number of object samples; translating each object sample in the large number of object samples; repeating the above steps until the samples converge; determining the object samples that converge to the same point as object samples of the same cluster to obtain multiple clusters; and calculating the cluster center corresponding to each cluster in the multiple clusters to obtain multiple cluster centers.

[0149] In some embodiments of this specification, the first type of attribute feature data includes at least one of the following data: customer object data, usage channel data, processing procedure data, transaction transparency data, and transaction nature data.

[0150] In some embodiments of this specification, the second type of attribute feature data includes at least one of the following: customer number data, high-risk customer number data, number of transactions, and transaction amount data.

[0151] As can be seen from the above description, the embodiments of this specification achieve the following technical effects: By assessing the risk of the target object from both the first type of attribute feature data and the second type of attribute feature data, the features contained in the target object data are made more comprehensive, thus improving the accuracy of risk identification. Furthermore, by using the Naive Bayes decision algorithm to complete or correct the second type of attribute feature data in the target object data, the completeness and correctness of the data in product classification can be guaranteed. Moreover, the Naive Bayes decision algorithm is more efficient and faster than other algorithms. After obtaining complete and correct product data, cluster analysis can be performed on the product data to obtain the risk category corresponding to the target object, which can improve the accuracy of target object risk identification, enhance the security of product transactions, protect user rights, and improve user experience.

[0152] This specification also provides a computer device, which can be found in the following description. Figure 6 The diagram shown illustrates the computer device structure based on the target object risk identification method provided in the embodiments of this specification. Specifically, the computer device may include an input device 61, a memory 62, and a processor 63. The memory 62 stores processor-executable instructions. When the processor 63 executes the instructions, it implements the steps of the target object risk identification method described in any of the above embodiments.

[0153] In this embodiment, the input device can specifically be one of the main devices for information exchange between the user and the computer system. The input device may include a keyboard, mouse, camera, scanner, light pen, handwriting input tablet, voice input device, etc.; the input device is used to input raw data and programs for processing these data into the computer. The input device can also receive data transmitted from other modules, units, and devices. The processor can be implemented in any suitable manner. For example, the processor can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers, etc. The memory can specifically be a memory device used to store information in modern information technology. The memory can include multiple layers; in digital systems, anything that can store binary data can be considered memory; in integrated circuits, a circuit without physical form but with storage function is also called memory, such as RAM, FIFO, etc.; in a system, a storage device with physical form is also called memory, such as a memory stick, TF card, etc.

[0154] In this embodiment, the specific functions and effects implemented by the computer device can be explained in comparison with other embodiments, and will not be repeated here.

[0155] This specification also provides a computer storage medium based on a target object risk identification method, wherein the computer storage medium stores computer program instructions that, when executed, implement the steps of the target object risk identification method described in any of the above embodiments.

[0156] In this embodiment, the storage medium includes, but is not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), cache, hard disk drive (HDD), or memory card. The memory can be used to store computer program instructions. The network communication unit can be an interface configured according to standards specified in the communication protocol for network connection communication.

[0157] In this embodiment, the specific functions and effects implemented by the program instructions stored in the computer storage medium can be explained by comparison with other embodiments, and will not be repeated here.

[0158] Obviously, those skilled in the art will understand that the modules or steps of the embodiments described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the embodiments of this specification are not limited to any particular combination of hardware and software.

[0159] It should be understood that the above description is for illustrative purposes and not for limitation. Many embodiments and applications beyond the provided examples will be apparent to those skilled in the art upon reading the above description. Therefore, the scope of this specification should not be determined by reference to the above description, but rather by reference to the foregoing claims and the full scope of their equivalents.

[0160] The above description is merely a preferred embodiment of this specification and is not intended to limit this specification. Various modifications and variations can be made to the embodiments described herein by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this specification should be included within the scope of protection of this specification.

Claims

1. A method for identifying the risk of a target object, characterized in that, include: Receive risk detection requests; The risk detection request carries the target object identifier; In response to the risk detection request, target object data corresponding to the target object identifier is obtained; the target object data includes a first type of attribute feature data and a second type of attribute feature data; the first type of attribute feature data is used to characterize the business attribute features associated with the target object; the second type of attribute feature data is used to characterize the transaction attribute features associated with the target object; In the case where there are missing and / or incorrect indicator data in the second type of attribute feature data, the Naive Bayes decision algorithm is used to complete the missing indicator data and / or correct the incorrect indicator data in the second type of attribute feature data, so as to obtain the completed and / or corrected second type of attribute feature data. Cluster analysis is performed on the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data to obtain the risk category corresponding to the target object identifier; The step of using the Naive Bayes decision algorithm to complete missing indicator data and / or correct erroneous indicator data in the second type of attribute feature data includes: The first type of attribute feature data and the second type of attribute feature data are preprocessed to obtain preprocessed first type of attribute feature data and preprocessed second type of attribute feature data; The second type of attribute feature data is inspected to determine whether there are missing or incorrect indicator data in the second type of attribute feature data; If it is determined that there are missing and / or incorrect indicator data in the preprocessed second type of attribute feature data, the Naive Bayes decision algorithm is used to complete the missing indicator data and / or correct the incorrect indicator data in the second type of attribute feature data. The preprocessed second-type attribute feature data includes multiple second-type attribute features, including second-type attribute features with known and correct values ​​as well as second-type attribute features with unknown or incorrect values. Accordingly, the Naive Bayes decision algorithm is used to complete the missing indicator data and / or correct erroneous indicator data in the second type of attribute feature data, including: The Naive Bayes decision algorithm is used to calculate the conditional probability of different values ​​for the unknown or incorrect second-type attribute features, given that the values ​​of the second-type attribute features are known and correct. Based on the value of the second type of attribute feature with the highest conditional probability that is unknown or incorrect, data completion is performed on the missing indicator data and / or the incorrect indicator data is corrected in the second type of attribute feature data.

2. The target object risk identification method according to claim 1, characterized in that, Cluster analysis is performed on the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data to obtain the risk category corresponding to the target object identifier, including: Feature extraction is performed on the first type of attribute feature data and the padded and / or corrected second type of attribute feature data to obtain the target feature vector corresponding to the target object identifier; Calculate the distance between the target feature vector and the feature vector of the cluster center corresponding to each risk category in multiple risk categories; The risk category corresponding to the cluster center with the smallest distance from the target feature vector is determined as the risk category corresponding to the target object identifier.

3. The target object risk identification method according to claim 2, characterized in that, Also includes: Obtain the feature vector corresponding to each object sample from a large number of object samples; the large number of object samples includes object samples with known risk categories; Cluster analysis is performed on the feature vectors corresponding to each object sample in the large number of object samples to obtain multiple cluster centers; Calculate the distance between the feature vector of the object sample of the known risk category and each cluster center in the plurality of cluster centers to determine the risk category corresponding to each cluster center in the plurality of cluster centers.

4. The target object risk identification method according to claim 3, characterized in that, Cluster analysis was performed on the feature vectors corresponding to each object sample in the large number of object samples to obtain multiple cluster centers, including: From the large number of object samples, the feature vectors corresponding to each object sample in multiple object samples are randomly selected as the initial cluster centers to obtain multiple cluster centers; Repeat the following steps until the feature vectors corresponding to the multiple cluster centers no longer change: calculate the distance between the feature vectors corresponding to other object samples in a large number of object samples besides the object samples corresponding to the multiple cluster centers and the feature vectors corresponding to each cluster center in the multiple cluster centers, and assign them to the nearest cluster center to obtain multiple clusters; calculate the cluster center of each cluster in the multiple clusters.

5. The target object risk identification method according to claim 3, characterized in that, Cluster analysis was performed on the feature vectors corresponding to each object sample in the large number of object samples to obtain multiple cluster centers, including: Calculate the average displacement of each object sample in the large number of object samples; Each object sample in the large number of object samples is translated; Repeat the above steps until the samples converge. Object samples that converge to the same point are identified as object samples of the same cluster, resulting in multiple clusters. Calculate the cluster center corresponding to each of the multiple clusters to obtain multiple cluster centers.

6. The target object risk identification method according to claim 1, characterized in that, The first type of attribute feature data includes at least one of the following: Customer data, usage channel data, processing procedure data, transaction transparency data, and transaction nature data.

7. The target object risk identification method according to claim 1, characterized in that, The second type of attribute feature data includes at least one of the following: Customer number data, high-risk customer number data, number of transactions, and transaction amount data.

8. A target object risk identification device, characterized in that, include: The receiving module is used to receive risk detection requests; The risk detection request carries the target object identifier; The acquisition module is used to acquire target object data corresponding to the target object identifier in response to the risk detection request; the target object data includes a first type of attribute feature data and a second type of attribute feature data; the first type of attribute feature data is used to characterize the business attribute features associated with the target object; the second type of attribute feature data is used to characterize the transaction attribute features associated with the target object; The data completion and correction module is used to complete the missing indicator data and / or correct the erroneous indicator data in the second type of attribute feature data when there are missing indicator data and / or erroneous indicator data in the second type of attribute feature data, so as to obtain the completed and / or corrected second type of attribute feature data. The clustering analysis module is used to perform clustering analysis on the first type of attribute feature data and the supplemented and / or corrected second type of attribute feature data to obtain the risk category corresponding to the target object identifier; The completion and correction module is specifically used for: The first type of attribute feature data and the second type of attribute feature data are preprocessed to obtain preprocessed first type of attribute feature data and preprocessed second type of attribute feature data; The second type of attribute feature data is inspected to determine whether there are missing or incorrect indicator data in the second type of attribute feature data; If it is determined that there are missing and / or incorrect indicator data in the preprocessed second type of attribute feature data, the Naive Bayes decision algorithm is used to complete the missing indicator data and / or correct the incorrect indicator data in the second type of attribute feature data. The preprocessed second-type attribute feature data includes multiple second-type attribute features, including second-type attribute features with known and correct values ​​as well as second-type attribute features with unknown or incorrect values. Accordingly, the completion and correction module is specifically used for: The Naive Bayes decision algorithm is used to calculate the conditional probability of different values ​​for the unknown or incorrect second-type attribute features, given that the values ​​of the second-type attribute features are known and correct. Based on the value of the second type of attribute feature with the highest conditional probability that is unknown or incorrect, data completion is performed on the missing indicator data and / or the incorrect indicator data is corrected in the second type of attribute feature data.

9. A computer device, characterized in that, It includes a processor and a memory for storing processor-executable instructions, wherein the processor, when executing the instructions, implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 7.

Images