Method, apparatus, electronic device, and storage medium for data processing

Abstract

The application discloses a data processing method and device, electronic equipment and storage medium, and relates to the technical field of data processing. A specific embodiment of the method comprises: receiving a processing instruction of business risk, obtaining business data of each business party, extracting a label feature set from the business data according to each preset label; calculating the importance of each label feature in the label feature set, and screening each target feature according to the importance; determining each target feature combination and the corresponding risk level according to the association between each target feature and each business party; and constructing a risk portrait model according to the target feature combination and the risk level to perform risk assessment on each business party. The embodiment can solve the problem that the participation of business experience in the screening feature is relatively high, which not only leads to low efficiency of feature screening, but also leads to reduced accuracy of the constructed risk portrait system.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a data processing method, apparatus, electronic device, and storage medium. Background Technology

[0002] In various business scenarios, to improve the quality and security of business processes, a risk profiling system can be built for business stakeholders to assess and manage risks. For example, in the weighing and verification business of logistics, a risk profiling system can be built to detect instances of weight theft or underweighting. Related technologies typically integrate behavioral and operational data from business processes to develop multiple features, which are then manually selected based on business experience to construct the risk profiling system. However, this feature selection method relies heavily on business experience. When faced with a large number of features, this not only leads to low efficiency in feature selection but also reduces the accuracy of the resulting risk profiling system. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a data processing method, apparatus, electronic device, and storage medium that can solve the problem that the high degree of involvement of business experience in the method of feature screening not only leads to low efficiency in feature screening but also reduces the accuracy of the constructed risk profile system.

[0004] To achieve the above objectives, according to one aspect of the present invention, a data processing method is provided.

[0005] A data processing method according to an embodiment of the present invention includes: receiving a business risk processing instruction, obtaining business data from each business party, and extracting a set of tag features from the business data according to each preset tag;

[0006] Calculate the importance of each label feature in the label feature set, and filter each target feature according to the importance.

[0007] The combination of target features and the corresponding risk level are determined based on the relationship between each target feature and each business entity.

[0008] A risk profile model is constructed based on the combination of target features and the risk level to conduct risk assessments for each of the business parties.

[0009] In one embodiment, calculating the importance of each label feature in the label feature set includes:

[0010] A preset construction model is invoked to construct multiple feature decision trees based on each label feature in the label feature set;

[0011] For each of the label features, the Gini impurity of the label feature when it is a split node in each of the feature decision trees is calculated, so that the statistical value of each Gini impurity is determined as the importance of the label feature.

[0012] In yet another embodiment, calculating the Gini impurity of the label feature as a splitting node in each of the feature decision trees includes:

[0013] For each of the feature decision trees, the first Gini impurity of the label feature before it becomes a split node and the second Gini impurity after it becomes a split node are calculated, and the difference between the first Gini impurity and the second Gini impurity is determined as the Gini impurity of the label feature corresponding to the feature decision tree.

[0014] In yet another embodiment, calculating the importance of each label feature in the label feature set includes:

[0015] Calculate the probability that each label feature in the label feature set does not belong to the same category, so as to determine the statistical value of the probability corresponding to each label feature as the corresponding importance.

[0016] In another embodiment, determining each target feature combination and its corresponding risk level based on the association between each target feature and each user includes:

[0017] Combine the target features and calculate the frequency of each feature combination based on the relationship between each target feature and each business party in order to filter the target feature combinations;

[0018] The risk level is determined based on the frequency of each combination of target features.

[0019] In another embodiment, the target features are combined, and the frequency of each feature combination is calculated based on the association between each target feature and each business party to filter the target feature combinations, including:

[0020] The first frequency of each target feature is calculated based on the relationship between each target feature and each business party, so as to filter each target feature;

[0021] The selected target features are combined to obtain feature combinations. The second frequency of each feature combination is calculated based on the relationship between each feature combination and each business party, so as to filter the feature combinations.

[0022] The selected feature combinations and the selected target features are determined as the target feature combinations.

[0023] To achieve the above objectives, according to another aspect of the present invention, a data processing apparatus is provided.

[0024] An embodiment of the present invention provides a data processing apparatus comprising: a receiving unit, configured to receive a processing instruction for business risks, acquire business data from each business party, and extract a set of tag features from the business data according to each preset tag;

[0025] A filtering unit is used to calculate the importance of each label feature in the label feature set and filter each target feature according to the importance.

[0026] The determining unit is used to determine the combination of target features and the corresponding risk level based on the relationship between each target feature and each business party;

[0027] A construction unit is used to construct a risk model based on the target feature combination and the risk level, so as to conduct risk assessment for each of the business parties.

[0028] In one embodiment, the filtering unit is specifically used for:

[0029] A preset construction model is invoked to construct multiple feature decision trees based on each label feature in the label feature set;

[0030] For each of the label features, the Gini impurity of the label feature when it is a split node in each of the feature decision trees is calculated, so that the statistical value of each Gini impurity is determined as the importance of the label feature.

[0031] In yet another embodiment, the filtering unit is specifically used for:

[0032] Calculate the probability that each label feature in the label feature set does not belong to the same category, so as to determine the statistical value of the probability corresponding to each label feature as the corresponding importance.

[0033] In yet another embodiment, the filtering unit is specifically used for:

[0034] For each of the feature decision trees, the first Gini impurity of the label feature before it becomes a split node and the second Gini impurity after it becomes a split node are calculated, and the difference between the first Gini impurity and the second Gini impurity is determined as the Gini impurity of the label feature corresponding to the feature decision tree.

[0035] In yet another embodiment, the determining unit is specifically used for:

[0036] Combine the target features and calculate the frequency of each feature combination based on the relationship between each target feature and each business party in order to filter the target feature combinations;

[0037] The risk level is determined based on the frequency of each combination of target features.

[0038] In yet another embodiment, the determining unit is specifically used for:

[0039] The first frequency of each target feature is calculated based on the relationship between each target feature and each business party, so as to filter each target feature;

[0040] The selected target features are combined to obtain feature combinations. The second frequency of each feature combination is calculated based on the relationship between each feature combination and each business party, so as to filter the feature combinations.

[0041] The selected feature combinations and the selected target features are determined as the target feature combinations.

[0042] To achieve the above objectives, according to another aspect of the present invention, an electronic device is provided.

[0043] An electronic device according to an embodiment of the present invention includes: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the data processing method provided in the embodiment of the present invention.

[0044] To achieve the above objectives, according to another aspect of the present invention, a computer-readable medium is provided.

[0045] An embodiment of the present invention provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the data processing method provided in the embodiment of the present invention.

[0046] To achieve the above objectives, according to another aspect of the present invention, a computer program product is provided.

[0047] A computer program product according to an embodiment of the present invention includes a computer program that, when executed by a processor, implements the data processing method provided in the embodiment of the present invention.

[0048] One embodiment of the above invention has the following advantages or beneficial effects:

[0049] In this embodiment of the invention, after receiving a business risk processing instruction, business data from each business party can be obtained. A set of tag features can be extracted based on preset tags. Then, the importance of each tag feature in the set is calculated, and target features are filtered. Based on the correlation between each target feature and each business party, the combination of target features and the corresponding risk level can be determined. Furthermore, a risk model is constructed based on the combination of target features and the risk level. Thus, in this embodiment of the invention, after extracting tag features from business data, feature filtering can be performed based on two dimensions: the importance of the tag features and their correlation with the business party. This achieves multi-dimensional feature extraction and filtering, and determines the corresponding risk level for constructing a risk profile model. This avoids feature filtering based on business experience, improving the efficiency of feature filtering and the accuracy of the constructed risk profile system.

[0050] The further effects of the aforementioned unconventional alternative methods will be explained below in conjunction with specific implementation methods. Attached Figure Description

[0051] The accompanying drawings are provided to better understand the invention and are not intended to unduly limit the scope of the invention. Wherein:

[0052] Figure 1 This is a schematic diagram of a main flow of a data processing method according to an embodiment of the present invention;

[0053] Figure 2 This is a schematic diagram of another main flow of a data processing method according to an embodiment of the present invention;

[0054] Figure 3 This is a schematic diagram of the main units of a data processing apparatus according to an embodiment of the present invention;

[0055] Figure 4 This is an exemplary system architecture diagram in which embodiments of the present invention can be applied;

[0056] Figure 5 This is a schematic diagram of the structure of a computer system suitable for implementing embodiments of the present invention. Detailed Implementation

[0057] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention, including various details to aid understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0058] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined with each other. The acquisition, transmission, storage, use, and processing of data in this application comply with relevant national laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0059] This invention provides a data processing system that can be used for business data processing scenarios, specifically for feature processing of business data in the construction of a risk profiling system. For example, taking the weighing business in a logistics scenario, constructing a corresponding risk profiling system typically requires based on business characteristics. Therefore, it is necessary to extract and filter features from the business data to construct a risk profiling model. This risk profiling model can then be used to implement a risk profiling system, which in turn allows for risk assessment of various business parties.

[0060] This invention provides a data processing method, which can be executed by a data processing system, such as... Figure 1 As shown, the method includes the following steps.

[0061] S101: Receives instructions for handling business risks, obtains business data from each business party, and extracts a set of tag features from the business data based on each preset tag.

[0062] The processing instruction refers to the instruction for constructing the risk profile model. The processing instruction may include information about the business type, such as a business type identifier, to facilitate the acquisition of business data. The processing instruction can be sent to an external system or automatically triggered by the system. For example, to improve the accuracy of the risk profile model, it needs to be updated periodically. Therefore, the data processing system can be set to periodically trigger processing instructions to periodically execute the method of this embodiment to update the risk profile model. The business party can refer to the business application party. For example, in a logistics scenario, the weighing party can include the merchant responsible for the logistics of goods, who performs the weighing business. Therefore, business data can be obtained from each business party in this step. The preset label can refer to a pre-set label for feature extraction. In this embodiment, a label system can be pre-established, and feature extraction can be performed based on the label system to obtain a label feature set.

[0063] Specifically, the preset tags in this embodiment of the invention can include various types, such as statistical tags and rule tags. Statistical tags can be further divided into attribute tags and behavioral tags. Generally, attribute tags are mostly discrete tags, while behavioral tags are mostly continuous tags. For example, taking the weighing business in a logistics scenario as an example, attribute tags can include Boolean tags (such as whether the business is a trusted object, whether it is a new registration, etc.) and categorized tags (such as the province or region to which the business belongs, etc.). Behavioral tags can include operational and business data (such as the business's order volume, freight, discounts, sampling volume, discrepancy orders, discrepancy rate, cooperation duration, etc.). Rule tags can represent tag information given based on business experience, such as active business (orders placed for more than 15 consecutive days within a month), high-frequency business partners, and high-frequency business sites.

[0064] It should be noted that, in this embodiment of the invention, the business data can be preprocessed before feature extraction, and after the features are extracted, the feature data can be discretized to facilitate data processing.

[0065] S102: Calculate the importance of each label feature in the label feature set, and filter each target feature according to the importance.

[0066] The importance of the label features can represent the importance of their impact on the business risk profile. In this step, the label feature set is filtered according to the importance to obtain the target features, which may include multiple features.

[0067] Specifically, in this step, the calculation of importance can be performed as follows: calculate the probability that each label feature in the label feature set does not belong to the same category, so as to determine the statistical value of the probability corresponding to each label feature as the corresponding importance.

[0068] For a set of label features, where the features are extracted from different labels, the correlation between the labels is not uniform. Therefore, the probability that each label feature does not belong to the same category can be calculated to represent the importance of the label. This step can calculate the probability of label features not belonging to the same category using a preset algorithm, such as a machine learning algorithm or a Bayesian algorithm. After calculating the probabilities for each label feature, statistical values ​​of these probabilities can be calculated to determine their corresponding importance. These statistical values ​​can include the mean, median, etc.

[0069] On the other hand, the calculation of importance in this step can be performed as follows: call the preset construction model to construct multiple feature decision trees based on each label feature in the label feature set; for each label feature, calculate the Gini impurity when the label feature is a split node in each feature decision tree, so as to determine the statistical value of each Gini impurity as the importance of the label feature.

[0070] In this step, data mining can be performed using the GBDT algorithm. The global importance of a feature is measured by its importance within the constructed decision tree, which is reflected in its Gini impurity. The model can be built based on the GBDT algorithm to construct multiple feature decision trees, with each label feature's position within the tree indicating different levels of importance. The Gini impurity of the label features is frequently calculated within the decision tree to determine whether a node is a splitting node. Therefore, in this step, the importance of the label features can be determined based on their Gini impurity when they act as splitting nodes. Statistical values ​​can be set according to requirements, such as the average, median, maximum, minimum, etc. Taking the average as an example, the formula for calculating the importance of the label features is shown in Formula 1.

[0071]

[0072] In Formula 1, imp i T represents the importance of the i-th label feature, M represents the number of feature decision trees, and T represents the importance of the i-th label feature. m Let imp represent the decision tree for the m-th feature. i (T m ) represents the Gini impurity of the decision tree corresponding to the m-th feature of the i-th label feature.

[0073] It should be noted that in the feature decision tree, whether the label feature is used as a split node is determined by the Gini impurity. Therefore, after determining that the label feature can be used as a split node to continue splitting, the Gini impurity will be calculated again to determine whether further splitting is needed. Thus, the Gini impurity changes before and after the label feature is used as a split node.

[0074] In this embodiment of the invention, the change value of Gini impurity can be determined as a reflection of importance. Therefore, the process of calculating the Gini impurity of the label feature corresponding to the feature decision tree in this step can be executed as follows: for each feature decision tree, calculate the first Gini impurity of the label feature before it becomes a split node and the second Gini impurity after it becomes a split node, and determine the difference between the first Gini impurity and the second Gini impurity as the Gini impurity of the label feature corresponding to the feature decision tree.

[0075] For each feature decision tree, the Gini impurity of the label feature that serves as a split node can be calculated before the split and after the split. That is, the first Gini impurity of the label feature before it serves as a split node and the second Gini impurity after it serves as a split node. Then, the difference between the two can be calculated. In this way, the Gini impurity of each label feature corresponding to each feature decision tree can be calculated.

[0076] For example, if label feature X1 is a split node in feature decision tree 1, with a first Gini impurity of a11 and a second Gini impurity of a12, the difference can be calculated as (a11-a12). If label feature X1 is a split node in feature decision tree 2, with a first Gini impurity of a21 and a second Gini impurity of a22, the difference can be calculated as (a21-a22). Similarly, if feature X1 is a split node in feature decision tree m, with a first Gini impurity of am1 and a second Gini impurity of am2, the difference can be calculated as (am1-am2). Thus, the importance of label feature X1 can be calculated as imp = ((a11-a12) + (a21-a22) + ... + (am1-am2)) / m.

[0077] In this embodiment of the invention, after calculating the importance of each tag feature, they can be sorted according to their importance to select a preset number of tag features as target features.

[0078] It should be noted that, in this embodiment of the invention, the calculation of importance in this step can be implemented by a classification model, such as by using a classifier model model = XGBClassifier(). Furthermore, a feature importance function model.feature_importances_ can be introduced to rank the importance of the label features, and the ranking can be visualized, such as by using plot_importance(model).

[0079] After constructing the classification model, it can be trained and its performance tested, and features can be selected. For example, the function `SelectFromModel(model, threshold=thresh, prefit=True)` can be called to progressively input features, and the accuracy_score function can be called to calculate the accuracy and judge the model's performance. Generally, model performance (such as model complexity and accuracy) decreases as the number of selected features increases. Therefore, in this embodiment of the invention, a balance can be struck between model complexity and accuracy, selecting a model that is easy to understand and has high accuracy.

[0080] S103: Determine the combination of target characteristics and the corresponding risk level based on the relationship between each target characteristic and each business party.

[0081] The correlation between target features and various business parties indicates the frequency of occurrence of each target feature in business processing. Therefore, target feature combinations are calculated based on these correlations to further filter features. Risk levels are pre-set; the risk level corresponding to each target feature combination is calculated to reflect the degree of risk of that combination.

[0082] Specifically, this step can be performed as follows: combine the target features, calculate the frequency of each feature combination based on the relationship between each target feature and each business party, and screen each target feature combination; determine the corresponding risk level based on the frequency of each target feature combination.

[0083] Each target feature can be arbitrarily combined, and the resulting feature combination may include one or more target features. Then, the frequency of each feature combination is calculated, and the risk level is determined based on the frequency.

[0084] Frequency can represent the number of related business parties. A feature combination associated with multiple business parties indicates a significant impact on business processing. Therefore, feature combinations with high frequency can be selected as target feature combinations. Risk levels can be pre-set, such as including high-risk, medium-risk, and low-risk feature combinations. In this step, the frequency range corresponding to each risk feature can be preset, allowing the determination of the corresponding risk level based on the frequency of each target feature combination.

[0085] In one implementation, the method for screening target feature combinations in this step can be specifically executed as follows: calculate the first frequency of each target feature based on the association between each target feature and each business party, so as to screen each target feature; combine the screened target features to obtain each feature combination; calculate the second frequency of each feature combination based on the association between each feature combination and each business party, so as to screen the feature combination; and determine the screened feature combination and the screened target features as the target feature combination.

[0086] For each target feature, the filtering process can be performed sequentially according to the number of target features included in the feature combination. For example, the first frequency corresponding to a single target feature can be calculated for filtering to obtain the filtered target features. Then, the filtered target features can be combined to calculate the second frequency corresponding to each feature combination, and further filtering of feature combinations can be performed. This can reduce the filtering process and computational load.

[0087] For example, taking business parties including T1, T2, T3, and T4, and target features including A, B, C, D, and E as an example, the relationship between each target feature and the business parties can be shown in Table 1.

[0088] Table 1

[0089] Business side Target features T1 A, C, D T2 B, C, E T3 A, B, C, E T4 B, E

[0090] As shown in Table 1, the frequency of each target feature can be calculated: the frequency of A is 50%, the frequency of B is 75%, the frequency of C is 75%, the frequency of D is 25%, and the frequency of E is 75%. In this embodiment of the invention, the minimum support can be set to 50% for filtering, and the filtered target features are: A, B, C, and E.

[0091] Furthermore, A, B, C, and E can be paired and their corresponding frequencies calculated according to Table 1. The results are as follows: the frequency of {A, B} is 25%, the frequency of {A, C} is 50%, the frequency of {A, E} is 25%, the frequency of {B, C} is 50%, the frequency of {B, E} is 75%, and the frequency of {C, E} is 50%. Based on the minimum support, the target feature combinations are: {A, C}, {B, C}, {B, E}, {C, E}.

[0092] Furthermore, the above target feature combinations can be combined pairwise again to obtain feature combinations including three target features. The frequency of each combination is calculated according to Table 1, yielding the following results: the frequency of {A, B, C} is 25%, the frequency of {B, C, E} is 50%, and the frequency of {A, C, E} is 25%. Based on minimum support, the target feature combination is determined to be {B, C, E}.

[0093] Through the iterative process described above, the selected target feature combinations {{A}, {B}, {C}, {E}, {A, C}, {B, C}, {B, E}, {C, E}, {B, C, E}} can be obtained.

[0094] It should be noted that, in this embodiment of the invention, the continuous features can be discretized with equal width before performing this step, so that each feature belongs to a discrete value.

[0095] S104: Construct a risk profile model based on the combination of target features and risk levels to conduct risk assessments for each business unit.

[0096] In this embodiment of the invention, the method of constructing the risk profile model is not limited after obtaining the target feature combination and risk level. After the risk profile model is constructed, it can be used to conduct risk assessment for each business party.

[0097] It should be noted that, in order to improve the accuracy of the risk profile model in this embodiment of the invention, the risk profile model can be updated periodically or according to a preset cycle. Specifically, the selected features can be updated and enriched through this embodiment of the invention, thereby reconstructing the risk profile model and updating the risk profile model.

[0098] In this embodiment of the invention, after extracting tag features from business data, feature filtering can be performed based on two dimensions: the importance of the tag features and their association with the business party. The corresponding risk level can then be determined for use in constructing a risk profile model. This avoids feature filtering based on business experience, thereby improving the efficiency of feature filtering and the accuracy of the constructed risk profile system.

[0099] The following is combined Figure 1 The illustrated embodiments provide a detailed description of the data processing methods in the embodiments of the present invention, such as... Figure 2 As shown, the method includes the following steps.

[0100] S201: Receive the business risk processing instruction, obtain the business data of each business party, and extract the tag feature set from the business data according to each preset tag.

[0101] S202: Call the preset construction model to construct multiple feature decision trees based on each label feature in the label feature set.

[0102] S203: For each label feature, calculate the Gini impurity when the label feature is a split node in each feature decision tree, so as to determine the statistical value of each Gini impurity as the importance of the label feature.

[0103] S204: Select target features from the label feature set according to their importance.

[0104] S205: Combine the target features and calculate the frequency of each feature combination based on the relationship between each target feature and each business party in order to filter the target feature combinations.

[0105] S206: Determine the corresponding risk level based on the frequency of combinations of target features.

[0106] S207: Construct a risk profile model based on the combination of target characteristics and risk levels to conduct risk assessments for each business unit.

[0107] It should be noted that the data processing principle in the embodiments of the present invention is the same as... Figure 1 The data processing principles in the illustrated embodiments are the same and will not be repeated here.

[0108] To address the problems existing in the prior art, embodiments of the present invention provide a data processing apparatus 300, such as... Figure 3 As shown, the device 300 includes:

[0109] The receiving unit 301 is used to receive business risk processing instructions, obtain business data from each business party, and extract a set of tag features from the business data according to each preset tag.

[0110] The filtering unit 302 is used to calculate the importance of each label feature in the label feature set and filter each target feature according to the importance.

[0111] The determining unit 303 is used to determine the combination of target features and the corresponding risk level based on the relationship between each target feature and each business party;

[0112] Construction unit 304 is used to construct a risk model based on the target feature combination and the risk level, so as to conduct risk assessment for each of the business parties.

[0113] It should be understood that the manner in which embodiments of the present invention are implemented is different from the implementation method. Figure 1 The embodiments shown are the same and will not be described again here.

[0114] In one embodiment, the filtering unit 302 is specifically used for:

[0115] A preset construction model is invoked to construct multiple feature decision trees based on each label feature in the label feature set;

[0116] For each of the label features, the Gini impurity of the label feature when it is a split node in each of the feature decision trees is calculated, so that the statistical value of each Gini impurity is determined as the importance of the label feature.

[0117] In yet another embodiment, the filtering unit 302 is specifically used for:

[0118] Calculate the probability that each label feature in the label feature set does not belong to the same category, so as to determine the statistical value of the probability corresponding to each label feature as the corresponding importance.

[0119] In yet another embodiment, the filtering unit 302 is specifically used for:

[0120] For each feature decision tree, calculate the first Gini impurity of the label feature before it becomes a split node and the second Gini impurity after it becomes a split node. The difference between the first Gini impurity and the second Gini impurity is determined as the Gini impurity of the label feature corresponding to the feature decision tree.

[0121] In yet another embodiment, the determining unit 303 is specifically used for:

[0122] Combine the target features and calculate the frequency of each feature combination based on the relationship between each target feature and each business party in order to filter the target feature combinations;

[0123] The risk level is determined based on the frequency of each combination of target features.

[0124] In yet another embodiment, the determining unit 303 is specifically used for:

[0125] The first frequency of each target feature is calculated based on the relationship between each target feature and each business party, so as to filter each target feature;

[0126] The selected target features are combined to obtain feature combinations. The second frequency of each feature combination is calculated based on the relationship between each feature combination and each business party, so as to filter the feature combinations.

[0127] The selected feature combinations and the selected target features are determined as the target feature combinations.

[0128] It should be understood that the manner in which embodiments of the present invention are implemented is different from the implementation method. Figure 1 , 2 The embodiments shown are the same and will not be described again here.

[0129] In this embodiment of the invention, after extracting tag features from business data, feature filtering can be performed based on two dimensions: the importance of the tag features and their association with the business party. The corresponding risk level can then be determined for use in constructing a risk profile model. This avoids feature filtering based on business experience, thereby improving the efficiency of feature filtering and the accuracy of the constructed risk profile system.

[0130] According to embodiments of the present invention, an electronic device and a readable storage medium are also provided.

[0131] An electronic device according to an embodiment of the present invention includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform the data processing method provided in the embodiment of the present invention.

[0132] Figure 4 An exemplary system architecture 400 is shown, which can be applied to a data processing method or a data processing apparatus according to embodiments of the present invention.

[0133] like Figure 4 As shown, system architecture 400 may include terminal devices 401, 402, and 403, a network 404, and a server 405. Network 404 serves as the medium for providing communication links between terminal devices 401, 402, and 403 and server 405. Network 404 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0134] Users can use terminal devices 401, 402, and 403 to interact with server 405 via network 404 to receive or send messages, etc. Various client applications can be installed on terminal devices 401, 402, and 403.

[0135] Terminal devices 401, 402, and 403 can be various electronic devices with displays that support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0136] Server 405 can be a server that provides various services. The server can analyze and process data such as received product information query requests, and feed back the processing results (such as product information - just an example) to the terminal device.

[0137] It should be noted that the data processing method provided in the embodiments of the present invention is generally executed by server 405, and correspondingly, the data processing device is generally disposed in server 405.

[0138] It should be understood that Figure 4 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0139] The following is for reference. Figure 5 It shows a schematic diagram of the structure of a computer system 500 suitable for implementing embodiments of the present invention. Figure 5 The computer system shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0140] like Figure 5 As shown, the computer system 500 includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 502 or programs loaded from storage section 508 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the system 500. The CPU 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0141] The following components are connected to I / O interface 505: an input section 506 including a keyboard, mouse, etc.; an output section 507 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to I / O interface 505 as needed. A removable medium 511, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 510 as needed so that computer programs read from it can be installed into storage section 508 as needed.

[0142] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 509, and / or installed from removable medium 511. When the computer program is executed by central processing unit (CPU) 501, it performs the functions defined above in the system of this invention.

[0143] It should be noted that the computer-readable medium shown in this invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0144] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a unit, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0145] The units described in the embodiments of the present invention can be implemented in software or hardware. The described units can also be housed in a processor; for example, a processor can be described as including a receiving unit, a filtering unit, a determining unit, and a constructing unit. The names of these units do not necessarily limit the specific unit; for example, a receiving unit can also be described as a "unit that functions as a receiving unit."

[0146] In another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiments; or it may exist independently and not assembled into the device. The computer-readable medium carries one or more programs that, when executed by the device, cause the device to perform the data processing method provided by the present invention.

[0147] In another aspect, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the data processing method provided in the embodiments of the present invention.

[0148] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A data processing method, characterized in that, include: Receive instructions for handling business risks, acquire business data from each business party, and extract a set of tag features from the business data based on each preset tag; Calculate the importance of each label feature in the label feature set, and filter each target feature according to the importance. The combination of target features and the corresponding risk level are determined based on the relationship between each target feature and each business entity. A risk profile model is constructed based on the combination of target features and the risk level to conduct risk assessments for each of the business parties.

2. The method according to claim 1, characterized in that, Calculating the importance of each label feature in the label feature set includes: A preset construction model is invoked to construct multiple feature decision trees based on each label feature in the label feature set; For each of the label features, the Gini impurity of the label feature when it is a split node in each of the feature decision trees is calculated, so that the statistical value of each Gini impurity is determined as the importance of the label feature.

3. The method according to claim 2, characterized in that, Calculating the Gini impurity of the label feature as a splitting node in each feature decision tree includes: For each of the feature decision trees, the first Gini impurity of the label feature before it becomes a split node and the second Gini impurity after it becomes a split node are calculated, and the difference between the first Gini impurity and the second Gini impurity is determined as the Gini impurity of the label feature corresponding to the feature decision tree.

4. The method according to claim 1, characterized in that, Calculating the importance of each label feature in the label feature set includes: Calculate the probability that each label feature in the label feature set does not belong to the same category, so as to determine the statistical value of the probability corresponding to each label feature as the corresponding importance.

5. The method according to claim 1, characterized in that, Determine each combination of target features and its corresponding risk level based on the association between each target feature and each user, including: Combine the target features and calculate the frequency of each feature combination based on the relationship between each target feature and each business party in order to filter the target feature combinations; The risk level is determined based on the frequency of each combination of target features.

6. The method according to claim 5, characterized in that, Combining the target features, calculating the frequency of each feature combination based on the association between each target feature and each business party, and filtering each target feature combination, including: The first frequency of each target feature is calculated based on the relationship between each target feature and each business party, so as to filter each target feature; The selected target features are combined to obtain feature combinations. The second frequency of each feature combination is calculated based on the relationship between each feature combination and each business party, so as to filter the feature combinations. The selected feature combinations and the selected target features are determined as the target feature combinations.

7. A data processing apparatus, characterized in that, include: The receiving unit is used to receive business risk processing instructions, obtain business data from each business party, and extract a set of tag features from the business data according to each preset tag. A filtering unit is used to calculate the importance of each label feature in the label feature set and filter each target feature according to the importance. The determining unit is used to determine the combination of target features and the corresponding risk level based on the relationship between each target feature and each business party; A construction unit is used to construct a risk model based on the target feature combination and the risk level, so as to conduct risk assessment for each of the business parties.

8. The apparatus according to claim 7, characterized in that, The filtering unit is specifically used for: A preset construction model is invoked to construct multiple feature decision trees based on each label feature in the label feature set; For each of the label features, the Gini impurity of the label feature when it is a split node in each of the feature decision trees is calculated, so that the statistical value of each Gini impurity is determined as the importance of the label feature.

9. An electronic device, characterized in that, include: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-6.

10. A computer-readable medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-6.

Images