Classification methods, devices, electronic devices and storage media of interactive objects

Abstract

This application discloses a method, apparatus, electronic device, and storage medium for classifying interactive objects. The method involves: acquiring at least one identical interactive feature of the interactive objects in an information system; dividing a preset feature value interval corresponding to the interactive feature based on a target feature value segmentation point to obtain sub-feature value intervals for the interactive feature; for each sub-feature value interval, calculating the proportion of interactive objects belonging to a preset type based on the feature value of the interactive feature and the sub-feature value interval, obtaining the type object proportion corresponding to the sub-feature value interval; calculating the information entropy of the interactive feature relative to the preset type at the target feature value segmentation point based on the type object proportions corresponding to each sub-feature value interval; selecting a target interactive feature from the interactive features based on the information entropy of each interactive feature relative to the preset type; and classifying the interactive objects to be processed based on the target interactive feature to obtain the classification result. This application can improve the accuracy and efficiency of interactive object classification.

Description

Technical Field

[0001] This application relates to the field of computer technology, and more specifically to a method, apparatus, electronic device, and storage medium for classifying interactive objects. Background Technology

[0002] With the development of the internet, the amount of data that online businesses (such as classification processing) need to process has increased dramatically. This data often has high feature dimensionality, making direct processing computationally extremely expensive. Therefore, it is necessary to select features, filtering out those that are not very helpful for classification processing and retaining those that are closely related to it.

[0003] In current related technologies, features are generally selected manually, which requires a lot of manpower and resources, and the feature selection efficiency is relatively low, which is not conducive to improving the accuracy and efficiency of classification. Summary of the Invention

[0004] This application provides a method, apparatus, electronic device, and storage medium for classifying interactive objects, which can improve the accuracy and efficiency of classifying interactive objects.

[0005] This application provides a method for classifying interactive objects, including:

[0006] Obtain feature information of at least one interactive object, the feature information including at least one identical interactive feature of the interactive object in the information system;

[0007] The interaction feature is segmented based on the target feature value segmentation point to obtain at least two sub-feature value intervals of the interaction feature.

[0008] For each sub-feature value interval, the proportion of interactive objects belonging to a preset type is calculated based on the feature value of the interactive feature and the sub-feature value interval, thus obtaining the proportion of type objects corresponding to the sub-feature value interval;

[0009] Based on the proportion of type objects corresponding to each sub-feature value interval, the type information entropy of the interaction feature is calculated to obtain the information entropy of the interaction feature to the preset type at the target feature value segmentation point;

[0010] Based on the information entropy of each interaction feature to a preset type, a target interaction feature is selected from the interaction features;

[0011] The interaction objects to be processed are classified according to the target interaction features to obtain the classification results.

[0012] Accordingly, embodiments of this application provide a classification device for interactive objects, including:

[0013] An acquisition unit is used to acquire feature information of at least one interactive object, the feature information including at least one identical interactive feature of the interactive object in the information system;

[0014] The segmentation unit is used to segment the preset feature value interval corresponding to the interactive feature based on the target feature value segmentation point to obtain at least two sub-feature value intervals of the interactive feature.

[0015] The calculation unit is used to calculate the proportion of interactive objects belonging to a preset type for each sub-feature value interval, based on the feature value of the interactive feature and the sub-feature value interval, so as to obtain the proportion of type objects corresponding to the sub-feature value interval.

[0016] The entropy calculation unit is used to calculate the type information entropy of the interaction feature based on the proportion of type objects corresponding to each sub-feature value interval, so as to obtain the information entropy of the interaction feature to the preset type under the target feature value segmentation point;

[0017] The selection unit is used to select a target interaction feature from the interaction features based on the information entropy of each interaction feature on a preset type.

[0018] The classification unit is used to classify the interactive object to be processed according to the target interactive features and obtain the classification result.

[0019] Optionally, in some embodiments of this application, the calculation unit may include a statistical subunit and a calculation subunit, as follows:

[0020] The statistical subunit is used to count the interactive objects whose feature values ​​of the interaction feature fall into the sub-feature value interval for each sub-feature value interval, so as to obtain the target interactive object corresponding to the sub-feature value interval.

[0021] The calculation subunit is used to calculate the proportion of the target interactive object whose type belongs to a preset type, and to obtain the proportion of type objects corresponding to the sub-feature value interval.

[0022] Optionally, in some embodiments of this application, the selection unit may include a determining subunit and a feature selection subunit, as follows:

[0023] The determining subunit is used to determine the target information entropy of the interaction feature pair of the preset type based on the information entropy of the interaction feature pair of the preset type under each target feature value segmentation point.

[0024] The feature selection subunit is used to select target interaction features from the interaction features based on the target information entropy of each interaction feature of a preset type.

[0025] Optionally, in some embodiments of this application, the determining subunit may be specifically used to sort the information entropy of the interaction feature pairs of a preset type under each target feature value segmentation point to obtain sorted information entropy; and select the target information entropy of the interaction feature pairs of a preset type from the sorted information entropy according to a predetermined rule.

[0026] Optionally, in some embodiments of this application, the determining subunit may be specifically used to determine the weight corresponding to the information entropy of the interaction feature to the preset type at each target feature value segmentation point; based on the weight, the information entropy of the interaction feature to the preset type at each target feature value segmentation point is fused to obtain the target information entropy of the interaction feature to the preset type.

[0027] Optionally, in some embodiments of this application, the entropy calculation unit may include an entropy calculation subunit and a fusion subunit, as follows:

[0028] The entropy calculation subunit is used to calculate the type information entropy of the sub-feature value interval according to the proportion of type objects corresponding to the sub-feature value interval, so as to obtain the sub-information entropy of the sub-feature value interval for a preset type.

[0029] The fusion subunit is used to fuse the sub-information entropy of each sub-feature value interval to a preset type, so as to obtain the information entropy of the interaction feature to the preset type under the target feature value segmentation point.

[0030] Optionally, in some embodiments of this application, the entropy calculation subunit may be used to perform a logarithmic operation on the proportion of type objects corresponding to the sub-feature value interval to obtain the logarithmic operation result of the sub-feature value interval; and to fuse the proportion of type objects corresponding to the sub-feature value interval and the logarithmic operation result to obtain the sub-information entropy of the sub-feature value interval for a preset type.

[0031] Optionally, in some embodiments of this application, the selection unit may be specifically used to select target interactive features from the interactive features whose information entropy for a preset type is less than a preset information entropy.

[0032] Optionally, in some embodiments of this application, the acquisition unit may include an acquisition subunit, an interaction subunit, and a selection subunit, as follows:

[0033] The acquisition subunit is used to acquire initial feature information of at least one interactive object, the initial feature information including at least one initial interaction feature of the interactive object in the information system;

[0034] An interaction subunit is used to perform feature interaction processing on the initial interaction features to obtain at least one candidate interaction feature of the interaction object.

[0035] A sub-unit is selected to select common interaction features of each interaction object from the candidate interaction features, so as to obtain feature information of at least one interaction object.

[0036] Optionally, in some embodiments of this application, the classification unit may include a feature acquisition subunit, a construction subunit, and a classification subunit, as follows:

[0037] The feature acquisition subunit is used to acquire the feature values ​​of the interactive object to be processed under the target interactive feature;

[0038] A construction subunit is used to perform feature construction processing on the interaction object to be processed based on the feature value of the interaction object under the target interaction feature, so as to obtain the object feature of the interaction object to be processed;

[0039] The classification subunit is used to classify the interactive object to be processed according to the object characteristics and obtain the classification result.

[0040] An electronic device provided in this application includes a processor and a memory. The memory stores multiple instructions, and the processor loads the instructions to execute the steps in the interactive object classification method provided in this application.

[0041] Furthermore, embodiments of this application also provide a storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements the steps in the classification method for interactive objects provided in embodiments of this application.

[0042] This application provides a method, apparatus, electronic device, and storage medium for classifying interactive objects. It can acquire feature information of at least one interactive object, including at least one identical interactive feature of the interactive object in an information system. The method involves segmenting a preset feature value interval corresponding to the interactive feature based on a target feature value segmentation point, obtaining at least two sub-feature value intervals for the interactive feature. For each sub-feature value interval, based on the feature value of the interactive feature and the sub-feature value interval, the method calculates the proportion of interactive objects belonging to a preset type, obtaining the type object proportion corresponding to the sub-feature value interval. Based on the type object proportions corresponding to each sub-feature value interval, the method calculates the type information entropy of the interactive feature relative to the preset type at the target feature value segmentation point, obtaining the information entropy of the interactive feature relative to the preset type at the target feature value segmentation point, selecting a target interactive feature from the interactive features based on the information entropy of each interactive feature relative to the preset type, and classifying the interactive objects to be processed based on the target interactive feature to obtain a classification result. This application can improve the accuracy and efficiency of interactive object classification while reducing the consumption of human and material resources. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1a This is a schematic diagram of a scenario illustrating the classification method for interactive objects provided in an embodiment of this application;

[0045] Figure 1b This is a flowchart of the classification method for interactive objects provided in the embodiments of this application;

[0046] Figure 2 This is another flowchart of the classification method for interactive objects provided in the embodiments of this application;

[0047] Figure 3a This is a schematic diagram of the structure of the interactive object classification device provided in the embodiments of this application;

[0048] Figure 3b This is another structural schematic diagram of the classification device for interactive objects provided in the embodiments of this application;

[0049] Figure 3c This is another schematic diagram of the structure of the classification device for interactive objects provided in the embodiments of this application;

[0050] Figure 3d This is another structural schematic diagram of the classification device for interactive objects provided in the embodiments of this application;

[0051] Figure 3e This is another schematic diagram of the structure of the classification device for interactive objects provided in the embodiments of this application;

[0052] Figure 3f This is another schematic diagram of the structure of the classification device for interactive objects provided in the embodiments of this application;

[0053] Figure 4 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0054] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0055] This application provides a method, apparatus, electronic device, and storage medium for classifying interactive objects. Specifically, the object classification apparatus can be integrated into an electronic device, such as a terminal or server.

[0056] It is understood that the method for classifying interactive objects in this embodiment can be executed on a terminal, on a server, or jointly by a terminal and a server. The above examples should not be construed as limiting this application.

[0057] like Figure 1a As shown, an example of a classification method for interactive objects jointly executed by a terminal and a server is presented. The interactive object classification system provided in this application includes a terminal 10 and a server 11, etc.; the terminal 10 and the server 11 are connected via a network, such as a wired or wireless network, etc., wherein the interactive object classification device can be integrated into the server.

[0058] Server 11 can be used to: acquire feature information of at least one interactive object, the feature information including at least one identical interactive feature of the interactive object in the information system; divide the preset feature value interval corresponding to the interactive feature based on a target feature value segmentation point to obtain at least two sub-feature value intervals of the interactive feature; for each sub-feature value interval, calculate the proportion of the interactive object belonging to a preset type based on the feature value of the interactive feature and the sub-feature value interval to obtain the type object proportion corresponding to the sub-feature value interval; calculate the type information entropy of the interactive feature based on the type object proportions corresponding to each sub-feature value interval to obtain the information entropy of the interactive feature relative to the preset type at the target feature value segmentation point; select a target interactive feature from the interactive features based on the information entropy of each interactive feature relative to the preset type; and classify the interactive object to be processed based on the target interactive feature to obtain a classification result. Server 11 can be a single server, a server cluster composed of multiple servers, or a cloud server. In the interactive object classification method or apparatus disclosed in this application, multiple servers can form a blockchain, and the server is a node on the blockchain.

[0059] The terminal 10 can send feature information of at least one interactive object to the server 11, and can also receive classification results output by the server 11. The terminal 10 can include a mobile phone, smart TV, tablet computer, laptop computer, or personal computer (PC), etc. A client can also be set on the terminal 10, which can be an application client or a browser client, etc.

[0060] The steps for classifying the interactive objects of server 11 can also be performed by terminal 10.

[0061] The method for classifying interactive objects provided in this application relates to machine learning in the field of artificial intelligence. This method can improve the accuracy and efficiency of interactive object classification.

[0062] Artificial intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to have perception, reasoning, and decision-making capabilities. AI technology is a comprehensive discipline involving a wide range of fields, encompassing both hardware and software technologies. AI software technologies mainly include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0063] Machine learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory, among others. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instructional learning.

[0064] The following sections provide detailed descriptions of each example. It should be noted that the order in which the embodiments are described is not intended to limit the preferred order of the embodiments.

[0065] This embodiment will be described from the perspective of a classification device for interactive objects. This classification device for interactive objects can be integrated into an electronic device, such as a server or a terminal.

[0066] The interactive object classification method of this application embodiment can be applied to various classification scenarios. For example, the interactive object classification method provided in this embodiment can be used to quickly and accurately select the interactive features (specifically transaction features) of merchants, which helps to improve the accuracy of interactive object classification.

[0067] like Figure 1b As shown, the specific process of the classification method for this interactive object can be as follows:

[0068] 101. Obtain feature information of at least one interactive object, the feature information including at least one identical interactive feature of the interactive object in the information system.

[0069] In this context, the interaction object can refer to the subject or entity engaging in the interaction; for example, the interaction object could be a transaction object, such as a merchant. The information system is a system used to process interaction-related transactions, which can include refunds, settlements, and other transaction-related matters.

[0070] The feature information may include interactive features such as order anomaly rate, order complaint rate, transaction success rate, transaction activity, transaction history duration, transaction volume within a preset time period, and / or transaction amount within a preset time period. This embodiment does not limit these features.

[0071] The order anomaly rate is calculated as the ratio of the number of anomaly orders to the total number of orders. The order complaint rate is calculated as the ratio of the number of orders with complaints to the total number of orders. The transaction success rate is calculated as the ratio of the number of orders for which the merchant received payment to the total number of orders. Transaction activity is calculated as the ratio of the number of units of time during which transactions occurred within a preset time period to the total number of units of time contained within that preset time period (e.g., the number of days with transactions within 30 days divided by 30, where one day is considered one unit of time within the preset time period). Transaction history duration is calculated as the length of time from the time of the first transaction to the current time. Transaction volume within a preset time period is calculated as the number of orders within that preset time period. Transaction value within a preset time period is calculated as the total value of orders within that preset time period.

[0072] For an interactive object, it may have multi-dimensional interactive features. In this embodiment, the feature information may include the interactive features common to interactive objects in the information system. For example, the common interactive features (i.e. the same interactive features) may be the order anomaly rate of each interactive object.

[0073] Optionally, in some embodiments, the feature information of the interactive object can be obtained by preprocessing the initial feature information (i.e., raw data) of the interactive object. Through preprocessing, the initial feature information can be transformed into feature information with obvious physical or statistical significance. This preprocessing process can include missing value handling, binning, one-hot encoding, feature hashing, embedding, log transformation, feature scaling, normalization, and feature interaction, etc.

[0074] Missing value handling can involve directly deleting features with excessive missing values, or filling in missing values, such as with 0 or the feature's average value. Binning involves classifying data according to certain rules, specifically discretizing continuous variables; the discretized features can then be one-hot encoded based on their group. One-hot encoding, or one-bit valid encoding, uses an N-bit state register to encode N states, each state having its own independent register bit, and at any given time, only one bit is valid.

[0075] Feature hashing compresses high-dimensional feature vectors into lower-dimensional feature vectors while minimizing the loss of original features. Nested processing maps high-dimensional vectors to low-dimensional spaces; specifically, it can map large, sparse vectors to a low-dimensional space that preserves semantic relationships. For scenarios with highly differentiated data, taking the logarithm can alter the data distribution.

[0076] Feature scaling is used to standardize the range of data features, making the feature ranges comparable. Standardization can specifically be a non-linear feature transformation that converts the distribution of data to conform to a certain distribution (such as a normal distribution). It's important to note that standardization does not change the type of data distribution. For example, if the original data follows a normal distribution, it will still follow a normal distribution after standardization; only the mean and variance of the data will change.

[0077] Optionally, in this embodiment, the step "obtaining feature information of at least one interactive object" may include:

[0078] Obtain initial feature information of at least one interactive object, the initial feature information including at least one initial interaction feature of the interactive object in the information system;

[0079] The initial interaction features are subjected to feature interaction processing to obtain at least one candidate interaction feature of the interaction object;

[0080] Select common interaction features among the candidate interaction features to obtain feature information of at least one interaction object.

[0081] Initial feature information can also be considered as the raw data of the interactive object. Feature interaction processing can be used to construct features (i.e., feature derivation), specifically by performing feature statistical analysis on the initial interactive features across various dimensions. Through feature interaction processing, the feature information can contain statistically relevant features corresponding to the interactive information, such as count features, density features, and bias features. Feature statistical analysis can involve transformations or combinations such as addition, subtraction, multiplication, and division. After appropriate transformations or combinations, initial interactive features can possess strong informational value. For example, candidate feature information obtained through feature interaction could be a merchant's sales growth rate for the current month.

[0082] 102. Based on the target feature value segmentation point, the preset feature value interval corresponding to the interactive feature is segmented to obtain at least two sub-feature value intervals of the interactive feature.

[0083] The target feature value segmentation point is a feature value within a preset feature value interval corresponding to the interactive feature, which can be used to segment the preset feature value interval. The preset feature value interval can be set according to actual conditions, and this embodiment does not impose any restrictions on it. The target feature value segmentation point divides the preset feature value interval into at least two sub-feature value intervals.

[0084] In some embodiments, the target feature value segmentation point can be a segmentation point within a segmentation point set of the interactive features, and the segmentation point set can include multiple segmentation points. Specifically, each feature value in a preset feature value interval can be traversed, and each feature value in the preset feature value interval can be added to the segmentation point set of the interactive features. Specifically, the preset feature value interval can be a continuous interval; discrete points in the preset feature value interval can be taken as the target feature value segmentation point.

[0085] For example, the interaction feature is the order anomaly rate, and its preset feature value range can be set to 0 to 1. The target feature value segmentation point can be set to 0.1, 0.2, 0.3, ..., 0.8, or 0.9, etc. If the target feature value segmentation point is 0.4, then the two sub-feature value ranges obtained by segmentation are 0 to 0.4 and 0.4 to 1, respectively. In other embodiments, the segmentation point set of the interaction feature may include nine target feature value segmentation points: 0.1, 0.2, 0.3, ..., 0.8, and 0.9. For each target feature value segmentation point, the preset feature value range corresponding to the interaction feature is segmented based on the target feature value segmentation point to obtain two sub-feature value ranges.

[0086] 103. For each sub-feature value interval, calculate the proportion of the interactive object whose type belongs to a preset type based on the feature value of the interactive feature and the sub-feature value interval, and obtain the proportion of type objects corresponding to the sub-feature value interval.

[0087] The type can specifically be the tag information of the interactive object, such as the category to which the interactive object belongs. For example, the type can be the credit attribute of the merchant, and the type to which a certain interactive object belongs can be that the merchant belongs to a merchant with good credit.

[0088] Optionally, in this embodiment, the step "for each sub-feature value interval, calculate the proportion of interactive objects belonging to a preset type based on the feature value of the interactive feature and the sub-feature value interval, and obtain the proportion of type objects corresponding to the sub-feature value interval" may include:

[0089] For each sub-feature value interval, the interactive objects whose feature values ​​of the interaction features fall into the sub-feature value interval are counted to obtain the target interactive objects corresponding to the sub-feature value interval;

[0090] Calculate the proportion of the target interactive object whose type belongs to a preset type, and obtain the proportion of type objects corresponding to the sub-feature value interval.

[0091] For example, the type of an interactive object can include 0 and 1. Assuming the preset type is 1, the target feature value segmentation point of a certain interactive feature divides its preset feature value range into two sub-feature value ranges: greater than 10 and less than 10. For the sub-feature value range greater than 10, the type object ratio can be represented as the proportion of interactive objects of type 1 among interactive objects with feature values ​​greater than 10. For the sub-feature value range less than 10, the type object ratio can be represented as the proportion of interactive objects of type 1 among interactive objects with feature values ​​less than 10.

[0092] In one specific embodiment, the interaction feature is the order anomaly rate. The order anomaly rates for each interaction object are 0, 0.1, 0.2, 0.2, 0.3, 0.5, 0.6, and 0.7, respectively. The tag information (i.e., type) of interaction objects with order anomaly rates of 0.3, 0.6, and 0.7 is "poor credit," while the tag information of other interaction objects is "good credit." If the target feature value segmentation point is 0.4, then the preset feature value interval of 0 to 1 for this interaction feature can be divided into two sub-feature value intervals: 0 to 0.4 and 0.4 to 1. Within the sub-feature value range of 0 to 0.4, the interaction objects corresponding to the order anomaly rates (i.e., 0, 0.1, 0.2, 0.2, 0.3) falling within this sub-feature value range are taken as the target interaction objects for this sub-feature value range. Among these target interaction objects, the interaction object with an order anomaly rate of 0.3 has poor credit, while the other four target interaction objects have good credit. Assuming the preset type is good credit, the proportion of target interaction objects belonging to the preset type is 0.8 (i.e., 4 / 5), which means that the proportion of type objects corresponding to this sub-feature value range is 0.8.

[0093] Similarly, for the sub-feature value range of 0.4 to 1, the interaction objects corresponding to the order anomaly rates (i.e., 0.5, 0.6, and 0.7) falling within this sub-feature value range are taken as the target interaction objects for this sub-feature value range. Among these target interaction objects, the interaction objects with order anomaly rates of 0.6 and 0.7 have poor credit, while the target interaction objects with an order anomaly rate of 0.5 have good credit. Assuming the preset type is good credit, the proportion of target interaction objects belonging to the preset type is 1 / 3, which means that the proportion of type objects corresponding to this sub-feature value range is 1 / 3.

[0094] 104. Based on the proportion of type objects corresponding to each sub-feature value interval, calculate the type information entropy of the interaction feature to obtain the information entropy of the interaction feature to the preset type at the target feature value segmentation point.

[0095] Optionally, in this embodiment, the step "calculating the type information entropy of the interaction feature based on the proportion of type objects corresponding to each sub-feature value interval, and obtaining the information entropy of the interaction feature to the preset type at the target feature value segmentation point" may include:

[0096] Based on the proportion of type objects corresponding to the sub-feature value interval, type information entropy is calculated for the sub-feature value interval to obtain the sub-information entropy of the sub-feature value interval for the preset type;

[0097] The sub-information entropy of each sub-feature value interval to the preset type is fused to obtain the information entropy of the interactive feature to the preset type under the target feature value segmentation point.

[0098] There are various ways to fuse sub-information entropy, and this embodiment does not limit this one. For example, the fusion method can be weighted summation or mean calculation, etc.

[0099] Among them, the information entropy of the interaction feature with respect to the preset type represents the uncertainty of the interaction feature with respect to the preset type. Specifically, the lower the information entropy, the more effectively the interaction feature can determine the type of the interaction object. In other words, the stronger the interpretability of the interaction feature with respect to the preset type.

[0100] Specifically, the step "fusion of sub-information entropy of each sub-feature value interval to a preset type to obtain the information entropy of the interaction feature to the preset type under the target feature value segmentation point" may include:

[0101] Determine the weight of each sub-feature value interval for the sub-information entropy of a preset type;

[0102] Based on the weights, the sub-information entropy of each sub-feature value interval to the preset type is weighted and calculated to obtain the information entropy of the interaction feature to the preset type under the target feature value segmentation point.

[0103] The weights of each sub-feature value interval corresponding to the sub-information entropy of the preset type can be set according to the actual situation. This embodiment does not impose any restrictions on this, for example, they can all be set to 1.

[0104] In this embodiment, a target feature value segmentation point contains at least one feature value. That is, a target feature value segmentation point can contain one feature value or multiple feature values; this embodiment does not impose any restrictions on this. If a target feature value segmentation point contains only one feature value, the preset feature value interval can be divided into two sub-feature value intervals. If a target feature value segmentation point contains multiple (e.g., n) feature values, the preset feature value interval can be divided into n+1 sub-feature value intervals. It can be understood that the number of sub-feature value intervals can be the number of feature values ​​within the target feature value segmentation point plus one.

[0105] Optionally, in this embodiment, the step "calculating the type information entropy of the sub-feature value interval according to the proportion of type objects corresponding to the sub-feature value interval, and obtaining the sub-information entropy of the sub-feature value interval for a preset type" may include:

[0106] Perform a logarithmic operation on the proportion of type objects corresponding to the sub-feature value interval to obtain the logarithmic operation result of the sub-feature value interval;

[0107] The proportion of type objects corresponding to the sub-feature value interval and the logarithmic operation result are fused to obtain the sub-information entropy of the sub-feature value interval for the preset type.

[0108] Specifically, the logarithmic operation is the logarithmic operation. There are various ways to combine the proportion of type objects and the result of the logarithmic operation, and this embodiment does not limit this; for example, the combination method can be multiplication, etc.

[0109] Different target feature value segmentation points correspond to different sub-feature value intervals, and different sub-feature value intervals also have different sub-information entropies for the preset type. In some embodiments, the lower the sub-information entropy, the greater the difference in the proportion of type objects corresponding to the various sub-feature value intervals obtained by the target feature value segmentation point. That is, the target feature value segmentation point can more effectively distinguish the type of interactive object, making the sub-feature value intervals more interpretable of the type to which the interactive object belongs.

[0110] The step "merging the proportion of type objects corresponding to the sub-feature value interval and the logarithmic operation result to obtain the sub-information entropy of the sub-feature value interval for the preset type" may include:

[0111] Multiply the proportion of type objects corresponding to the sub-feature value interval by the logarithmic operation result to obtain the sub-information entropy of the sub-feature value interval for the preset type.

[0112] Specifically, in this embodiment, the calculation method for the information entropy of the interaction feature of the preset type can be shown in the following formula:

[0113]

[0114] Where U represents a certain interaction feature, E[.] represents the mean operation, and n is the number of sub-feature value intervals under a certain target feature value segmentation point. This represents the proportion of type objects corresponding to the i-th sub-feature value interval. This is the result of the logarithmic operation on the i-th sub-eigenvalue interval. That is, the information entropy of the interaction feature U on the preset type.

[0115] In other embodiments, the step "calculating the type information entropy of the sub-feature value interval based on the proportion of type objects corresponding to the sub-feature value interval, to obtain the sub-information entropy of the sub-feature value interval for a preset type" may include:

[0116] Perform a logarithmic operation on the proportion of type objects corresponding to the sub-feature value interval to obtain the sub-information entropy of the sub-feature value interval for the preset type.

[0117] Alternatively, the result of the above logarithmic operation can be directly used as the sub-information entropy of the sub-feature value interval for a preset type.

[0118] 105. Select target interaction features from the preset types based on the information entropy of each interaction feature.

[0119] In some embodiments, for each interactive feature, the target feature value segmentation point can be a segmentation point within a segmentation point set of the interactive feature. The segmentation point set can include at least one segmentation point. For each target feature value segmentation point within the segmentation point set, the above operation can be performed, specifically, calculating the information entropy of the interactive feature to a preset type under each target feature value segmentation point.

[0120] Optionally, in this embodiment, before the step "selecting target interaction features from the interaction features based on the information entropy of each interaction feature relative to a preset type", the following may be included:

[0121] Based on the information entropy of the interaction feature pair of the preset type at each target feature value segmentation point, determine the target information entropy of the interaction feature pair of the preset type.

[0122] The step "selecting target interaction features from the interaction features based on the information entropy of each interaction feature relative to a preset type" may include:

[0123] Based on the target information entropy of each interaction feature relative to a preset type, target interaction features are selected from the interaction features.

[0124] In this embodiment, for an interaction feature, its optimal target feature value segmentation point can also be determined. Specifically, for each target feature value segmentation point of the interaction feature, the information entropy of the interaction feature with respect to a preset type at that target feature value segmentation point can be calculated. In some embodiments, the target feature value segmentation point with the minimum information entropy can be taken as the optimal target feature value segmentation point, and this minimum information entropy is also the target information entropy of the interaction feature with respect to the preset type. After obtaining the target information entropy of each interaction feature with respect to the preset type, interaction features with target information entropy less than the preset information entropy can be selected as target interaction features.

[0125] Optionally, in this embodiment, the step "determining the target information entropy of the interaction feature pair of the preset type based on the information entropy of the interaction feature pair of the preset type under each target feature value segmentation point" may include:

[0126] The information entropy of the interaction features at each target feature value segmentation point is sorted according to a preset type to obtain the sorted information entropy.

[0127] According to predetermined rules, the target information entropy of the interaction feature pair of the preset type is selected from the sorted information entropy.

[0128] The predetermined rules can be set according to the actual situation, and this embodiment does not impose any restrictions on them.

[0129] In this embodiment, the information entropy of the interactive features at each target feature value segmentation point can be sorted from smallest to largest according to the size of the information entropy, or it can be sorted from largest to smallest. This embodiment does not limit this.

[0130] In one embodiment, if the information entropy is sorted from smallest to largest, the predetermined rule may be to use the information entropy of the first sorted information as the target information entropy of the interaction feature for the preset type. Alternatively, the predetermined rule may be to use the average of the information entropies of the first n sorted information as the target information entropy of the interaction feature for the preset type. This embodiment does not limit this.

[0131] Optionally, in this embodiment, the step "determining the target information entropy of the interaction feature pair of the preset type based on the information entropy of the interaction feature pair of the preset type under each target feature value segmentation point" may include:

[0132] Determine the weights of the interactive features corresponding to the information entropy of a preset type at each target feature value segmentation point;

[0133] Based on the weights, the information entropy of the interaction features at each target feature value segmentation point is fused to obtain the target information entropy of the interaction features at the preset type.

[0134] There are various ways to perform this fusion, such as weighted operation, and this embodiment does not limit this.

[0135] Optionally, in this embodiment, the step "selecting target interaction features from the interaction features based on the information entropy of each interaction feature relative to a preset type" may include:

[0136] Select target interaction features from the interaction features whose information entropy for a preset type is less than the preset information entropy.

[0137] The preset information entropy can be set according to the actual situation, and this embodiment does not impose any restrictions on it. For example, it can be set according to the number of target interaction features that need to be acquired.

[0138] Optionally, in other embodiments, the step "selecting target interaction features from the interaction features based on the information entropy of each interaction feature relative to a preset type" may include:

[0139] Based on the information entropy of the preset type, the interaction features are sorted.

[0140] Target interaction features are selected from the sorted interaction features based on preset rules.

[0141] The preset rules can be set according to the actual situation, and this embodiment does not impose any restrictions on them.

[0142] If the interaction features are sorted based on information entropy from smallest to largest, the preset rule can be to take the top n interaction features after sorting as the target interaction features. Here, n can be set according to the actual situation; this embodiment does not impose any restrictions on it. For example, it can be increased or decreased according to the computational load of the algorithm, such as being between 10 and 20.

[0143] This application can calculate the information entropy of each interaction feature with respect to the sample label (i.e., type). The lower the information entropy, the greater the internal variability of the interaction feature (specifically, the greater the difference in the distribution of feature values ​​of different types of interaction objects within the preset feature value range of the interaction feature; for example, interaction objects of type A are concentrated in the sub-feature value range where the feature value is less than 10, while interaction objects of type B are concentrated in the sub-feature value range where the feature value is greater than 10). This indicates that the interaction feature is more likely to be the point of difference between various types of interaction objects (i.e., samples). The lower the information entropy of the interaction feature, the stronger the explanatory power of the interaction feature for the target variable (i.e., sample label).

[0144] 106. Classify the interactive objects to be processed according to the target interactive features to obtain the classification results.

[0145] Optionally, in this embodiment, the step "classifying the interaction object to be processed according to the target interaction features and obtaining the classification result" may include:

[0146] Obtain the feature values ​​of the interactive object to be processed under the target interactive features;

[0147] Based on the feature values ​​of the interaction object to be processed under the target interaction feature, feature construction processing is performed on the interaction object to be processed to obtain the object feature of the interaction object to be processed.

[0148] The interactive objects to be processed are classified according to the object characteristics to obtain the classification results.

[0149] Specifically, the interactive objects to be processed can be classified according to their characteristics using a classifier. This classifier can be a Support Vector Machine (SVM), a Recurrent Neural Network (RNN), a Fully Connected Deep Neural Network (DNN), etc., and this embodiment does not impose any limitations on it.

[0150] This application can determine which interaction features are more accurate in predicting the type of an interactive object, thereby selecting target interaction features with a higher correlation to the preset type, i.e., target interaction features with stronger interpretability of the preset type, thus improving the accuracy of feature selection. This application can also reduce the consumption of manpower and resources, which is beneficial to improving the efficiency of feature selection. This application can be applied to merchant data modeling systems to improve the accuracy and efficiency of modeling, and can also be used in the classification of interactive objects to improve classification accuracy.

[0151] In the specific scenario of merchant transactions, the interaction objects are each merchant, and the type can be the merchant's credit attribute. The characteristic information of each merchant can be obtained. The characteristic information includes at least one common interaction feature of each merchant in the transaction system. The interaction feature can specifically include order abnormality rate, order complaint rate, transaction success rate, transaction activity, transaction history duration, transaction volume within a preset time period, and / or transaction amount within a preset time period, etc. Then, the preset feature value interval corresponding to the interaction feature is divided according to the target feature value segmentation point to obtain at least two sub-feature value intervals of the interaction feature.

[0152] For example, the interaction feature is the order anomaly rate, and its preset feature value range can be 0 to 1. The target feature value split point can be set to 0.1, 0.2, 0.3, ..., 0.8 or 0.9, etc. If the target feature value split point is 0.4, then the two sub-feature value ranges of the segmented order anomaly rate are 0 to 0.4 and 0.4 to 1, respectively.

[0153] Then, for each sub-feature value interval, the interaction objects whose feature values ​​fall within the sub-feature value interval are counted to obtain the target interaction objects corresponding to the sub-feature value interval; the proportion of target interaction objects belonging to a preset type is calculated to obtain the type object proportion corresponding to the sub-feature value interval. Specifically, the preset type can be merchants with good credit. For a sub-feature value interval of 0 to 0.4, the type object proportion can represent the proportion of merchants with a good credit type among those with an order anomaly rate of 0 to 0.4; for a sub-feature value interval of 0.4 to 1, the type object proportion can represent the proportion of merchants with a good credit type among those with an order anomaly rate of 0.4 to 1.

[0154] After obtaining the proportion of type objects corresponding to each sub-feature value interval, the type information entropy of the interaction feature can be calculated based on the proportion of type objects corresponding to each sub-feature value interval to obtain the information entropy of the interaction feature to the preset type at the target feature value segmentation point; and the target interaction feature can be selected from the interaction features based on the information entropy of each interaction feature to the preset type.

[0155] For example, in one embodiment, target interaction features that are important for determining a merchant's creditworthiness are selected from various interaction features of the merchant—such as order anomaly rate and transaction success rate. In practical applications, these selected target interaction features—order anomaly rate and transaction success rate—can be used in the merchant's credit classification. Specifically, the order anomaly rate and transaction success rate of the merchant to be classified can be obtained. For example, if the order anomaly rate is 0.01 and the transaction success rate is 0.9, feature construction processing is performed on the merchant to be classified to obtain the object features of the merchant to be classified. Based on these object features, it can be predicted whether the merchant to be classified is a merchant with good credit or a merchant with poor credit.

[0156] The interactive object classification method in this embodiment can select target interactive features from the interactive features that are more conducive to improving the classification accuracy of the interactive objects to be processed. Based on the feature values ​​of the interactive objects to be processed under the target interactive features, feature construction processing is performed, which can improve the classification accuracy of the interactive objects to be processed. In addition, feature construction processing is performed only based on the selected target interactive features, avoiding the application of all interactive features to the feature construction of the interactive objects to be processed, which greatly reduces the amount of computation and improves the efficiency of classification.

[0157] As can be seen from the above, this embodiment can obtain feature information of at least one interactive object, the feature information including at least one identical interactive feature of the interactive object in the information system; it divides the preset feature value interval corresponding to the interactive feature based on the target feature value segmentation point to obtain at least two sub-feature value intervals of the interactive feature; for each sub-feature value interval, it calculates the proportion of the interactive object belonging to a preset type according to the feature value of the interactive feature and the sub-feature value interval, obtaining the type object proportion corresponding to the sub-feature value interval; it calculates the type information entropy of the interactive feature according to the type object proportion corresponding to each sub-feature value interval, obtaining the information entropy of the interactive feature relative to the preset type under the target feature value segmentation point; it selects a target interactive feature from the interactive features according to the information entropy of each interactive feature relative to the preset type; and it classifies the interactive object to be processed according to the target interactive feature to obtain the classification result. This application selects features by using the information entropy of the interactive feature relative to the preset type, which can improve the accuracy of feature selection and reduce the consumption of manpower and material resources, thus improving the efficiency and accuracy of interactive object classification.

[0158] Based on the method described in the preceding embodiments, the following will provide a more detailed explanation by taking the specific integration of the classification device for the interactive object into the server as an example.

[0159] This application provides a method for classifying interactive objects, such as... Figure 2As shown, the specific process of the classification method for this interactive object can be as follows:

[0160] 201. The server obtains feature information of at least one interactive object, the feature information including at least one identical interactive feature of the interactive object in the information system.

[0161] In this context, the interaction object can refer to the subject or entity engaging in the interaction; for example, the interaction object could be a transaction object, such as a merchant. The information system is a system used to process interaction-related transactions, which can include refunds, settlements, and other transaction-related matters.

[0162] The feature information may include interactive features such as order anomaly rate, order complaint rate, transaction success rate, transaction activity, transaction history duration, transaction volume within a preset time period, and / or transaction amount within a preset time period. This embodiment does not limit these features.

[0163] 202. The server divides the preset feature value interval corresponding to the interactive feature based on the target feature value segmentation point to obtain at least two sub-feature value intervals of the interactive feature.

[0164] The target feature value segmentation point is a feature value within a preset feature value interval corresponding to the interactive feature, which can be used to segment the preset feature value interval. The preset feature value interval can be set according to actual conditions, and this embodiment does not impose any restrictions on it. The target feature value segmentation point divides the preset feature value interval into at least two sub-feature value intervals.

[0165] In some embodiments, the target feature value segmentation point can be a segmentation point within a segmentation point set of the interactive features, and the segmentation point set can include multiple segmentation points. Specifically, each feature value in a preset feature value interval can be traversed, and each feature value in the preset feature value interval can be added to the segmentation point set of the interactive features. Specifically, the preset feature value interval can be a continuous interval; discrete points in the preset feature value interval can be taken as the target feature value segmentation point.

[0166] 203. For each sub-feature value interval, the server counts the interactive objects whose feature values ​​of the interaction feature fall into the sub-feature value interval, and obtains the target interactive object corresponding to the sub-feature value interval.

[0167] 204. The server calculates the proportion of the target interactive object whose type belongs to a preset type, and obtains the proportion of type objects corresponding to the sub-feature value interval.

[0168] For example, the type of an interactive object can include 0 and 1. Assuming the preset type is 1, the target feature value segmentation point of a certain interactive feature divides its preset feature value range into two sub-feature value ranges: greater than 10 and less than 10. For the sub-feature value range greater than 10, the type object ratio can be represented as the proportion of interactive objects of type 1 among interactive objects with feature values ​​greater than 10. For the sub-feature value range less than 10, the type object ratio can be represented as the proportion of interactive objects of type 1 among interactive objects with feature values ​​less than 10.

[0169] 205. The server calculates the type information entropy of the interaction feature based on the proportion of type objects corresponding to each sub-feature value interval, and obtains the information entropy of the interaction feature to the preset type at the target feature value segmentation point.

[0170] Optionally, in this embodiment, the step "calculating the type information entropy of the interaction feature based on the proportion of type objects corresponding to each sub-feature value interval, and obtaining the information entropy of the interaction feature to the preset type at the target feature value segmentation point" may include:

[0171] Based on the proportion of type objects corresponding to the sub-feature value interval, type information entropy is calculated for the sub-feature value interval to obtain the sub-information entropy of the sub-feature value interval for the preset type;

[0172] The sub-information entropy of each sub-feature value interval to the preset type is fused to obtain the information entropy of the interactive feature to the preset type under the target feature value segmentation point.

[0173] There are various ways to fuse sub-information entropy, and this embodiment does not limit this one. For example, the fusion method can be weighted summation or mean calculation, etc.

[0174] Among them, the information entropy of the interaction feature with respect to the preset type represents the uncertainty of the interaction feature with respect to the preset type. Specifically, the lower the information entropy, the more effectively the interaction feature can determine the type of the interaction object. In other words, the stronger the interpretability of the interaction feature with respect to the preset type.

[0175] Optionally, in this embodiment, the step "calculating the type information entropy of the sub-feature value interval according to the proportion of type objects corresponding to the sub-feature value interval, and obtaining the sub-information entropy of the sub-feature value interval for a preset type" may include:

[0176] Perform a logarithmic operation on the proportion of type objects corresponding to the sub-feature value interval to obtain the logarithmic operation result of the sub-feature value interval;

[0177] The proportion of type objects corresponding to the sub-feature value interval and the logarithmic operation result are fused to obtain the sub-information entropy of the sub-feature value interval for the preset type.

[0178] Specifically, the logarithmic operation is the logarithmic operation. There are various ways to combine the proportion of type objects and the result of the logarithmic operation, and this embodiment does not limit this; for example, the combination method can be multiplication, etc.

[0179] 206. The server determines the target information entropy of the interaction feature pair of the preset type based on the information entropy of the interaction feature pair under each target feature value segmentation point.

[0180] Optionally, in this embodiment, the step "determining the target information entropy of the interaction feature pair of the preset type based on the information entropy of the interaction feature pair of the preset type under each target feature value segmentation point" may include:

[0181] The information entropy of the interaction features at each target feature value segmentation point is sorted according to a preset type to obtain the sorted information entropy.

[0182] According to predetermined rules, the target information entropy of the interaction feature pair of the preset type is selected from the sorted information entropy.

[0183] The predetermined rules can be set according to the actual situation, and this embodiment does not impose any restrictions on them.

[0184] In this embodiment, the information entropy of the interactive features at each target feature value segmentation point can be sorted from smallest to largest according to the size of the information entropy, or it can be sorted from largest to smallest. This embodiment does not limit this.

[0185] In one embodiment, if the information entropy is sorted from smallest to largest, the predetermined rule may be to use the information entropy of the first sorted information as the target information entropy of the interaction feature for the preset type. Alternatively, the predetermined rule may be to use the average of the information entropies of the first n sorted information as the target information entropy of the interaction feature for the preset type. This embodiment does not limit this.

[0186] 207. The server selects target interaction features from the preset type of target information entropy based on each interaction feature.

[0187] Optionally, in this embodiment, the step "selecting target interaction features from the interaction features based on the target information entropy of each interaction feature of a preset type" may include:

[0188] Select target interaction features from the interaction features whose target information entropy is less than the preset information entropy for the preset type.

[0189] The preset information entropy can be set according to the actual situation, and this embodiment does not impose any restrictions on it. For example, it can be set according to the number of target interaction features that need to be acquired.

[0190] 208. The server classifies the interactive objects to be processed based on the target interaction characteristics and obtains the classification results.

[0191] As can be seen from the above, this embodiment can obtain feature information of at least one interactive object through a server. The feature information includes at least one identical interactive feature of the interactive object in the information system. Based on the target feature value segmentation point, the preset feature value interval corresponding to the interactive feature is segmented to obtain at least two sub-feature value intervals of the interactive feature. For each sub-feature value interval, the interactive objects whose feature values ​​fall into the sub-feature value interval are counted to obtain the target interactive object corresponding to the sub-feature value interval. The proportion of the target interactive object whose type belongs to a preset type is calculated to obtain the type object proportion corresponding to the sub-feature value interval. Based on the type object proportion corresponding to each sub-feature value interval, the type information entropy of the interactive feature is calculated to obtain the information entropy of the interactive feature relative to the preset type at the target feature value segmentation point. Based on the information entropy of the interactive feature relative to the preset type at each target feature value segmentation point, the target information entropy of the interactive feature relative to the preset type is determined. Based on the target information entropy of each interactive feature relative to the preset type, a target interactive feature is selected from the interactive features. The interactive objects to be processed are classified according to the target interactive feature to obtain the classification result. Information entropy can characterize the uncertainty of information. This application selects features based on the information entropy of a preset type using interactive features. It can determine which interactive features can more accurately predict and determine the type of the interactive object, and select interactive features that are more correlated with the preset type, that is, interactive features that are more interpretable to the preset type, thereby improving the accuracy of feature selection and improving the classification accuracy of interactive objects. This application can also reduce the consumption of human and material resources and improve the classification efficiency of interactive objects.

[0192] To better implement the above methods, embodiments of this application also provide a classification device for interactive objects, such as... Figure 3a As shown, the classification device for interactive objects may include an acquisition unit 301, a segmentation unit 302, a calculation unit 303, an entropy calculation unit 304, a selection unit 305, and a classification unit 306, as follows:

[0193] (1) Obtain unit 301;

[0194] The acquisition unit 301 is used to acquire feature information of at least one interactive object, the feature information including at least one identical interactive feature of the interactive object in the information system.

[0195] Optionally, in some embodiments of this application, the acquisition unit 301 may include an acquisition subunit 3011, an interaction subunit 3012, and a selection subunit 3013, see [link to relevant documentation]. Figure 3b ,as follows:

[0196] The acquisition subunit 3011 is used to acquire initial feature information of at least one interactive object, the initial feature information including at least one initial interactive feature of the interactive object in the information system;

[0197] The interaction subunit 3012 is used to perform feature interaction processing on the initial interaction features to obtain at least one candidate interaction feature of the interaction object.

[0198] The selection subunit 3013 is used to select common interaction features of each interaction object from the candidate interaction features to obtain feature information of at least one interaction object.

[0199] (2) Segmentation unit 302;

[0200] The segmentation unit 302 is used to segment the preset feature value interval corresponding to the interactive feature based on the target feature value segmentation point to obtain at least two sub-feature value intervals of the interactive feature.

[0201] (3) Calculation unit 303;

[0202] The calculation unit 303 is used to calculate the proportion of interactive objects belonging to a preset type for each sub-feature value interval based on the feature value of the interactive feature and the sub-feature value interval, so as to obtain the proportion of type objects corresponding to the sub-feature value interval.

[0203] Optionally, in some embodiments of this application, the calculation unit 303 may include a statistical subunit 3031 and a calculation subunit 3032, see [link to relevant documentation]. Figure 3c ,as follows:

[0204] The statistical subunit 3031 is used to count the interactive objects whose feature values ​​of the interaction feature fall into the sub-feature value interval for each sub-feature value interval, so as to obtain the target interactive object corresponding to the sub-feature value interval.

[0205] The calculation subunit 3032 is used to calculate the proportion of the target interactive object whose type belongs to a preset type, and to obtain the proportion of type objects corresponding to the sub-feature value interval.

[0206] (4) Entropy calculation unit 304;

[0207] The entropy calculation unit 304 is used to calculate the type information entropy of the interactive feature based on the proportion of type objects corresponding to each sub-feature value interval, so as to obtain the information entropy of the interactive feature to the preset type under the target feature value segmentation point.

[0208] Optionally, in some embodiments of this application, the entropy calculation unit 304 may include an entropy calculation subunit 3041 and a fusion subunit 3042, see [link to relevant documentation]. Figure 3d ,as follows:

[0209] The entropy calculation subunit 3041 is used to calculate the type information entropy of the sub-feature value interval according to the proportion of type objects corresponding to the sub-feature value interval, so as to obtain the sub-information entropy of the sub-feature value interval for a preset type.

[0210] The fusion subunit 3042 is used to fuse the sub-information entropy of each sub-feature value interval to a preset type, so as to obtain the information entropy of the interaction feature to the preset type under the target feature value segmentation point.

[0211] Optionally, in some embodiments of this application, the entropy calculation subunit 3041 may be specifically used to perform a logarithmic operation on the proportion of type objects corresponding to the sub-feature value interval to obtain the logarithmic operation result of the sub-feature value interval; and to fuse the proportion of type objects corresponding to the sub-feature value interval and the logarithmic operation result to obtain the sub-information entropy of the sub-feature value interval for a preset type.

[0212] (5) Select unit 305;

[0213] The selection unit 305 is used to select a target interaction feature from the interaction features based on the information entropy of each interaction feature and a preset type.

[0214] Optionally, in some embodiments of this application, the selection unit 305 may include a determination subunit 3051 and a feature selection subunit 3052, see [link to relevant documentation]. Figure 3e ,as follows:

[0215] The determining subunit 3051 is used to determine the target information entropy of the interaction feature pair of the preset type based on the information entropy of the interaction feature pair of the preset type under each target feature value segmentation point.

[0216] The feature selection subunit 3052 is used to select target interaction features from the interaction features based on the target information entropy of the preset type of each interaction feature.

[0217] Optionally, in some embodiments of this application, the determining subunit 3051 may be used to sort the information entropy of the interaction feature pairs of a preset type under each target feature value segmentation point to obtain sorted information entropy; and select the target information entropy of the interaction feature pairs of a preset type from the sorted information entropy according to a predetermined rule.

[0218] Optionally, in some embodiments of this application, the determining subunit 3051 may be specifically used to determine the weight corresponding to the information entropy of the interaction feature to the preset type at each target feature value segmentation point; based on the weight, the information entropy of the interaction feature to the preset type at each target feature value segmentation point is fused to obtain the target information entropy of the interaction feature to the preset type.

[0219] Optionally, in some embodiments of this application, the selection unit 305 may be specifically used to select target interactive features from the interactive features whose information entropy for a preset type is less than a preset information entropy.

[0220] (6) Classification unit 306;

[0221] The classification unit 306 is used to classify the interactive object to be processed according to the target interactive features and obtain the classification result.

[0222] Optionally, in some embodiments of this application, the classification unit 306 may include a feature acquisition subunit 3061, a construction subunit 3062, and a classification subunit 3063, see [link to relevant documentation]. Figure 3f ,as follows:

[0223] The feature acquisition subunit 3061 is used to acquire the feature value of the interactive object to be processed under the target interactive feature;

[0224] The construction subunit 3062 is used to perform feature construction processing on the interaction object to be processed based on the feature value of the interaction object under the target interaction feature, so as to obtain the object feature of the interaction object to be processed.

[0225] The classification subunit 3063 is used to classify the interactive object to be processed according to the object characteristics and obtain the classification result.

[0226] As can be seen from the above, in this embodiment, the acquisition unit 301 acquires feature information of at least one interactive object, the feature information including at least one identical interactive feature of the interactive object in the information system; the segmentation unit 302 segments the preset feature value interval corresponding to the interactive feature based on the target feature value segmentation point to obtain at least two sub-feature value intervals of the interactive feature; the calculation unit 303 calculates the proportion of the interactive object belonging to a preset type based on the feature value of the interactive feature and the sub-feature value interval for each sub-feature value interval to obtain the type object proportion corresponding to the sub-feature value interval; the entropy calculation unit 304 calculates the type information entropy of the interactive feature based on the type object proportion corresponding to each sub-feature value interval to obtain the information entropy of the interactive feature relative to the preset type at the target feature value segmentation point; the selection unit 305 selects the target interactive feature from the interactive features based on the information entropy of each interactive feature relative to the preset type; and the classification unit 306 classifies the interactive object to be processed based on the target interactive feature to obtain the classification result. Information entropy can characterize the uncertainty of information. This application selects features based on the information entropy of a preset type using interactive features. It can determine which interactive features can more accurately predict and determine the type of the interactive object, and select interactive features that are more correlated with the preset type, that is, interactive features that are more interpretable to the preset type, thereby improving the accuracy of feature selection and improving the classification accuracy of interactive objects. This application can also reduce the consumption of human and material resources and improve the efficiency of classifying interactive objects.

[0227] This application also provides an electronic device, such as... Figure 4 The diagram shows a structural schematic of an electronic device involved in an embodiment of this application. This electronic device can be a terminal or a server, specifically:

[0228] The electronic device may include components such as a processor 401 with one or more processing cores, a memory 402 with one or more computer-readable storage media, a power supply 403, and an input unit 404. Those skilled in the art will understand that... Figure 4 The electronic device structure shown does not constitute a limitation on the electronic device and may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein:

[0229] The processor 401 is the control center of the electronic device. It connects various parts of the electronic device via various interfaces and lines. By running or executing software programs and / or modules stored in the memory 402, and by calling data stored in the memory 402, it performs various functions and processes data, thereby performing overall detection of the electronic device. Optionally, the processor 401 may include one or more processing cores; preferably, the processor 401 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 401.

[0230] The memory 402 can be used to store software programs and modules. The processor 401 executes various functional applications and data processing by running the software programs and modules stored in the memory 402. The memory 402 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 402 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory 402 may also include a memory controller to provide the processor 401 with access to the memory 402.

[0231] The electronic device also includes a power supply 403 that supplies power to the various components. Preferably, the power supply 403 can be logically connected to the processor 401 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 403 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0232] The electronic device may also include an input unit 404, which can be used to receive input digital or character information, and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0233] Although not shown, the electronic device may also include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 401 in the electronic device loads the executable files corresponding to the processes of one or more applications into the memory 402 according to the following instructions, and the processor 401 runs the applications stored in the memory 402 to realize various functions, as follows:

[0234] The process involves: acquiring feature information of at least one interactive object, including at least one identical interactive feature of the interactive object within the information system; segmenting the preset feature value interval corresponding to the interactive feature based on a target feature value segmentation point to obtain at least two sub-feature value intervals for the interactive feature; for each sub-feature value interval, calculating the proportion of interactive objects belonging to a preset type based on the feature value of the interactive feature and the sub-feature value interval, thus obtaining the type object proportion corresponding to the sub-feature value interval; calculating the type information entropy of the interactive feature based on the type object proportions corresponding to each sub-feature value interval, thus obtaining the information entropy of the interactive feature relative to the preset type at the target feature value segmentation point; selecting a target interactive feature from the interactive features based on the information entropy of each interactive feature relative to the preset type; and classifying the interactive objects to be processed based on the target interactive feature to obtain a classification result.

[0235] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0236] As can be seen from the above, this embodiment can obtain feature information of at least one interactive object, the feature information including at least one identical interactive feature of the interactive object in the information system; it divides the preset feature value interval corresponding to the interactive feature based on the target feature value segmentation point to obtain at least two sub-feature value intervals of the interactive feature; for each sub-feature value interval, it calculates the proportion of the interactive object belonging to a preset type according to the feature value of the interactive feature and the sub-feature value interval, obtaining the type object proportion corresponding to the sub-feature value interval; it calculates the type information entropy of the interactive feature according to the type object proportions corresponding to each sub-feature value interval, obtaining the information entropy of the interactive feature relative to the preset type at the target feature value segmentation point; it selects a target interactive feature from the interactive features according to the information entropy of each interactive feature relative to the preset type; and it classifies the interactive object to be processed according to the target interactive feature to obtain the classification result. This application uses the information entropy of the interactive feature relative to the preset type for feature selection, which can improve the accuracy of feature selection, reduce the consumption of manpower and material resources, and help improve the efficiency and accuracy of interactive object classification.

[0237] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0238] To this end, embodiments of this application provide a storage medium storing a plurality of instructions that can be loaded by a processor to execute steps in any of the interactive object classification methods provided in embodiments of this application. For example, the instructions can execute the following steps:

[0239] The process involves: acquiring feature information of at least one interactive object, including at least one identical interactive feature of the interactive object within the information system; segmenting the preset feature value interval corresponding to the interactive feature based on a target feature value segmentation point to obtain at least two sub-feature value intervals for the interactive feature; for each sub-feature value interval, calculating the proportion of interactive objects belonging to a preset type based on the feature value of the interactive feature and the sub-feature value interval, thus obtaining the type object proportion corresponding to the sub-feature value interval; calculating the type information entropy of the interactive feature based on the type object proportions corresponding to each sub-feature value interval, thus obtaining the information entropy of the interactive feature relative to the preset type at the target feature value segmentation point; selecting a target interactive feature from the interactive features based on the information entropy of each interactive feature relative to the preset type; and classifying the interactive objects to be processed based on the target interactive feature to obtain a classification result.

[0240] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0241] The storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0242] Since the instructions stored in the storage medium can execute the steps in any of the interactive object classification methods provided in the embodiments of this application, the beneficial effects that any of the interactive object classification methods provided in the embodiments of this application can achieve can be realized, as detailed in the preceding embodiments, and will not be repeated here.

[0243] According to one aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods provided in various alternative implementations of the aforementioned classification of interactive objects.

[0244] The foregoing has provided a detailed description of a method, apparatus, electronic device, and storage medium for classifying interactive objects according to embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method of classifying an interactive object, characterized by, The method comprises the following steps: obtaining feature information of at least one interactive object, the feature information comprising at least one same interactive feature of the interactive object in an information system; wherein the interactive object is a transaction object; the interactive feature comprises an order exception rate, an order complaint rate, a transaction success rate, a transaction activity level, a transaction history time length, a transaction volume in a preset time period, and / or a transaction amount in a preset time period; segmenting a preset feature value interval corresponding to the interactive feature based on a target feature value segmentation point, to obtain at least two sub-feature value intervals of the interactive feature; for each sub-feature value interval, calculating a proportion of types belonging to a preset type in the interactive object according to a feature value of the interactive feature and the sub-feature value interval, to obtain a type object proportion corresponding to the sub-feature value interval; performing type information entropy calculation on the interactive feature according to the type object proportion corresponding to each sub-feature value interval, to obtain information entropy of the preset type of the interactive feature under the target feature value segmentation point, comprising: performing type information entropy calculation on the sub-feature value interval according to the type object proportion corresponding to the sub-feature value interval, to obtain sub-information entropy of the preset type of the sub-feature value interval; and fusing the sub-information entropy of the preset type of each sub-feature value interval, to obtain information entropy of the preset type of the interactive feature under the target feature value segmentation point; selecting a target interactive feature from the interactive features according to the information entropy of the preset type of each interactive feature; classifying the target interactive feature according to the target interactive feature, to obtain a classification result.

2. The method of claim 1, wherein, The method further comprises the following steps before the step of selecting a target interactive feature from the interactive features according to the information entropy of the preset type of each interactive feature: for each sub-feature value interval, counting interactive objects whose feature values of the interactive feature fall into the sub-feature value interval, to obtain target interactive objects corresponding to the sub-feature value interval; calculating a proportion of types belonging to a preset type in the target interactive objects, to obtain a type object proportion corresponding to the sub-feature value interval.

3. The method of claim 1, wherein, The method further comprises the following steps before the step of selecting a target interactive feature from the interactive features according to the information entropy of the preset type of each interactive feature: determining a target information entropy of the preset type of the interactive feature based on the information entropy of the preset type of the interactive feature under each target feature value segmentation point; The method further comprises the following steps of selecting a target interactive feature from the interactive features according to the information entropy of the preset type of each interactive feature: selecting a target interactive feature from the interactive features according to the target information entropy of the preset type of each interactive feature.

4. The method of claim 3, wherein, The method further comprises the following steps of determining a target information entropy of the preset type of the interactive feature based on the information entropy of the preset type of the interactive feature under each target feature value segmentation point: sorting the information entropy of the preset type of the interactive feature under each target feature value segmentation point, to obtain sorted information entropy; selecting a target information entropy of the preset type of the interactive feature from the sorted information entropy according to a predetermined rule.

5. The method of claim 3, wherein, The target information entropy of the preset type of the interaction feature pair is determined based on the information entropy of the preset type of the interaction feature pair at each target feature value segmentation point, including: The weight corresponding to the information entropy of the preset type of the interaction feature pair at each target feature value segmentation point is determined; The information entropy of the preset type of the interaction feature pair at each target feature value segmentation point is fused based on the weight, to obtain the target information entropy of the preset type of the interaction feature pair.

6. The method of claim 1, wherein, The type information entropy calculation is performed on the sub-feature value interval according to the type object proportion corresponding to the sub-feature value interval, to obtain the sub-information entropy of the preset type of the sub-feature value interval, including: The logarithmic operation is performed on the type object proportion corresponding to the sub-feature value interval, to obtain the logarithmic operation result of the sub-feature value interval; The type object proportion corresponding to the sub-feature value interval and the logarithmic operation result are fused to obtain the sub-information entropy of the preset type of the sub-feature value interval.

7. The method of claim 1, wherein, The target interaction feature is selected from the interaction feature according to the information entropy of the preset type of each interaction feature, including: The target interaction feature with an information entropy less than a preset information entropy of the preset type is selected from the interaction feature.

8. The method of claim 1, wherein, The feature information of at least one interaction object is obtained, including: Initial feature information of at least one interaction object is obtained, the initial feature information including at least one initial interaction feature of the interaction object in an information system; The initial interaction feature is subjected to feature interaction processing to obtain at least one candidate interaction feature of the interaction object; Interaction features common to each interaction object are selected from the candidate interaction features to obtain the feature information of at least one interaction object.

9. The method of claim 1, wherein, The target interaction feature is selected from the interaction feature according to the information entropy of the preset type of each interaction feature, including: The feature value of a to-be-processed interaction object under the target interaction feature is obtained; The object feature of the to-be-processed interaction object is obtained by performing feature construction processing on the to-be-processed interaction object according to the feature value of the to-be-processed interaction object under the target interaction feature; The to-be-processed interaction object is classified according to the object feature to obtain a classification result.

10. A classification device for interactive objects, characterized in that, It includes: An acquisition unit is configured to acquire feature information of at least one interaction object, the feature information including at least one same interaction feature of the interaction object in an information system; wherein the interaction object is a transaction object; and the interaction feature includes an order abnormality rate, an order complaint rate, a transaction success rate, a transaction activity level, a transaction history time length, a transaction volume in a preset time period, and / or a transaction amount in a preset time period. A segmentation unit is configured to segment a preset feature value interval corresponding to the interaction feature based on a target feature value segmentation point, to obtain at least two sub-feature value intervals of the interaction feature. A calculation unit is configured to, for each sub-feature value interval, calculate a proportion of objects of a type belonging to a preset type in the interaction object according to a feature value of the interaction feature and the sub-feature value interval, to obtain a type object proportion corresponding to the sub-feature value interval. The entropy calculation unit is configured to calculate type information entropy of the interaction feature according to the type object proportion corresponding to each sub-feature value interval, and obtain information entropy of the interaction feature on a preset type at the target feature value segmentation point, including: calculating type information entropy of the sub-feature value interval according to the type object proportion corresponding to the sub-feature value interval, and obtaining sub-information entropy of the sub-feature value interval on the preset type; and fusing the sub-information entropy of each sub-feature value interval on the preset type to obtain the information entropy of the interaction feature on the preset type at the target feature value segmentation point. The selecting unit is configured to select a target interaction feature from the interaction features according to the information entropy of each interaction feature on the preset type. The classifying unit is configured to classify the to-be-processed interaction object according to the target interaction feature, and obtain a classification result.

11. The apparatus of claim 10, wherein, The calculation unit includes a statistical sub-unit and a calculation sub-unit, and the statistical sub-unit is configured to: for each sub-feature value interval, count interaction objects whose feature values of the interaction feature fall into the sub-feature value interval, and obtain target interaction objects corresponding to the sub-feature value interval. The calculation sub-unit is configured to calculate a proportion of type objects whose types belong to the preset type in the target interaction objects, and obtain the type object proportion corresponding to the sub-feature value interval. The selecting unit includes a determining sub-unit and a feature selecting sub-unit, and the determining sub-unit is configured to: determine target information entropy of the interaction feature on the preset type based on the information entropy of the interaction feature on the preset type at each target feature value segmentation point.

12. The apparatus of claim 10, wherein, The feature selecting sub-unit is configured to select a target interaction feature from the interaction features according to the target information entropy of each interaction feature on the preset type. The determining sub-unit is specifically configured to: sort the information entropy of the interaction feature on the preset type at each target feature value segmentation point to obtain sorted information entropy; and select the target information entropy of the interaction feature on the preset type from the sorted information entropy according to a predetermined rule. The determining sub-unit is specifically configured to: determine a weight corresponding to the information entropy of the interaction feature on the preset type at each target feature value segmentation point; and fuse the information entropy of the interaction feature on the preset type at each target feature value segmentation point based on the weight to obtain the target information entropy of the interaction feature on the preset type.

13. The apparatus of claim 12, wherein, The entropy calculation sub-unit is specifically configured to: perform logarithmic operation on the type object proportion corresponding to the sub-feature value interval to obtain a logarithmic operation result of the sub-feature value interval; and fuse the type object proportion corresponding to the sub-feature value interval and the logarithmic operation result to obtain the sub-information entropy of the sub-feature value interval on the preset type.

14. The apparatus of claim 12, wherein, The selecting unit is specifically configured to select a target interaction feature from the interaction features, the target interaction feature having information entropy on the preset type less than a preset information entropy.

15. The apparatus of claim 10, wherein, The obtaining unit includes an obtaining sub-unit, an interaction sub-unit and a selecting sub-unit, and the obtaining sub-unit is configured to: obtain initial feature information of at least one interaction object, the initial feature information including at least one initial interaction feature of the interaction object in an information system.

16. The apparatus of claim 10, wherein, ​ 17. The apparatus of claim 10, wherein, ​ ​ An interaction subunit is configured to perform feature interaction processing on the initial interaction feature to obtain at least one candidate interaction feature of the interaction object; A selection subunit is configured to select an interaction feature common to each interaction object from the candidate interaction features to obtain feature information of at least one interaction object.

18. The apparatus of claim 10, wherein, The classification unit includes a feature acquisition subunit, a construction subunit and a classification subunit, as follows: The feature acquisition subunit is configured to acquire a feature value of a to-be-processed interaction object under the target interaction feature; The construction subunit is configured to perform feature construction processing on the to-be-processed interaction object according to the feature value of the to-be-processed interaction object under the target interaction feature to obtain an object feature of the to-be-processed interaction object; The classification subunit is configured to classify the to-be-processed interaction object according to the object feature to obtain a classification result.

19. An electronic device, comprising: The memory stores an application program, and the processor is configured to run the application program in the memory to perform the operations in the classification method of the interaction object in any one of claims 1 to 9.

20. A storage medium, characterized by The storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by the processor to perform the steps in the classification method of the interaction object in any one of claims 1 to 9.

21. A computer program product, characterised in that, The computer program product includes computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer readable storage medium. The processor executes the computer instructions to cause the computer device to perform the steps in the classification method of the interaction object in any one of claims 1 to 9.

Images