A data processing method, apparatus, electronic device, and storage medium

Abstract

This disclosure provides a data processing method, apparatus, electronic device, and storage medium. The method includes: inputting a first row of data into a positive evaluation model to determine a first evaluation score for positively evaluating the recommendation effect; inputting a second row of data into a negative evaluation model to determine a second evaluation score for negatively evaluating the recommendation effect; and determining a final evaluation score based on the first and second evaluation scores. This disclosure combines the first evaluation score from the positive evaluation and the second evaluation score from the negative evaluation to assess the recommendation effect, resulting in a more comprehensive evaluation and better overall evaluation performance.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and more specifically, to a data processing method, apparatus, electronic device, and storage medium. Background Technology

[0002] With the rapid development of internet technology, massive amounts of data have been generated across various industries. How to recommend accurate and timely information to users from this vast amount of data has become a problem that recommendation systems need to solve.

[0003] Currently, various recommendation algorithms are emerging, but their recommendation effects vary. Taking user acquisition as an example, many recommendation algorithms can determine the number of new users to be recommended. However, a higher number of recommendations does not necessarily mean better recommendation results; that is, recommendation effectiveness cannot be determined solely from the number of recommendations. Therefore, evaluating recommendation effectiveness is fundamental to ensuring the effectiveness of recommendation algorithms. Summary of the Invention

[0004] This disclosure provides at least one data processing method, apparatus, electronic device, and storage medium to evaluate the recommendation effect from two levels, resulting in better performance evaluation.

[0005] In a first aspect, embodiments of this disclosure provide a data processing method, including:

[0006] Obtain the first and second behavioral data of at least one recommended user;

[0007] The first behavioral data is input into the positive evaluation model to determine a first evaluation score for positively evaluating the recommendation effect, and the second behavioral data is input into the negative evaluation model to determine a second evaluation score for negatively evaluating the recommendation effect.

[0008] The final evaluation score is determined based on the first evaluation score and the second evaluation score.

[0009] In one possible implementation, when there are multiple recommended users, determining the second evaluation score for reverse evaluation of the recommendation effect includes:

[0010] For the first recommended user among multiple recommended users, determine the second behavioral data of the first recommended user and the second behavioral data of the second recommended user associated with the first recommended user;

[0011] Input the second behavior data of the first recommended user and the second behavior data of the second recommended user into the reverse evaluation model to determine the risk assessment score corresponding to the first recommended user;

[0012] The risk assessment scores of multiple recommended users are aggregated to determine the second assessment score.

[0013] In one possible implementation, where the second behavioral data includes risk transaction data and risk attribute data, determining the risk assessment score corresponding to the first recommended user includes:

[0014] If a direct correlation is found between the first recommended user and the second recommended user based on the risk transaction data of each recommended user, the direct risk assessment sub-score corresponding to the first recommended user is determined.

[0015] Based on the risk attribute data of each recommended user, determine the indirect risk assessment sub-score between the first recommended user and other recommended users among the plurality of recommended users other than the first recommended user;

[0016] Based on the direct risk assessment sub-score and indirect risk assessment sub-score corresponding to the first recommended user, the risk assessment score corresponding to the first recommended user is determined.

[0017] In one possible implementation, determining the direct risk assessment sub-score corresponding to the first recommended user includes:

[0018] If it is determined that there is a direct association between the first recommended user and at least one second recommended user, a direct risk assessment sub-score for the first recommended user relative to each second recommended user is determined.

[0019] The direct risk assessment sub-score corresponding to the first recommended user is determined by summing the direct risk assessment sub-scores relative to at least one of the second recommended users.

[0020] In one possible implementation, when each of the second recommended users includes multiple direct risk assessment sub-scores, determining the direct risk assessment sub-score of the first recommended user relative to each of the second recommended users includes:

[0021] For each second recommended user, obtain multiple edge weight coefficients between the first recommended user and the second recommended user;

[0022] The direct risk assessment sub-score of the first recommended user relative to the second recommended user is determined by a weighted sum of multiple direct risk assessment sub-scores relative to the second recommended user and the edge weight coefficients corresponding to each direct risk assessment sub-score.

[0023] In one possible implementation, determining the indirect risk assessment sub-score between the first recommended user and other recommended users among the plurality of recommended users includes:

[0024] For the other recommended users among the multiple recommended users besides the first recommended user, the indirect risk assessment sub-score is determined based on the data similarity between the risk attribute data of the other recommended users and the risk attribute data of the first recommended user.

[0025] In one possible implementation, determining the risk assessment score corresponding to the first recommended user based on the direct risk assessment sub-score and the indirect risk assessment sub-score includes:

[0026] Based on the direct risk assessment sub-score corresponding to the first recommended user and the first assessment score of the first recommended user, the initial score of the first recommended user is determined;

[0027] The initial score is updated based on the indirect risk assessment sub-score to determine the risk assessment score corresponding to the first recommended user.

[0028] In one possible implementation, when there are multiple other recommended users, updating the initial score based on the indirect risk assessment sub-score to determine the risk assessment score corresponding to the first recommended user includes:

[0029] For each of the multiple other recommended users, the indirect risk assessment sub-score corresponding to the other recommended user and the initial score of the other recommended user are multiplied to determine the product result;

[0030] The initial score is updated by selecting the maximum value of the product result, and the risk assessment score corresponding to the first recommended user is determined.

[0031] In one possible implementation, when the first behavioral data includes multiple positive behavioral indicators, determining the first evaluation score for positively evaluating the recommendation effect includes:

[0032] For each of the at least one recommended user, obtain the indicator weight coefficient corresponding to each of the positive behavior indicators; and,

[0033] The multiple positive behavior indicators and the indicator weight coefficients corresponding to each positive behavior indicator are input into the positive evaluation model, and the first evaluation score for each recommended user is determined by weighted summation.

[0034] The first evaluation score determined for each recommended user is summed to determine the first evaluation score used to positively evaluate the recommendation effect.

[0035] In one possible implementation, determining the final evaluation score based on the first evaluation score and the second evaluation score includes:

[0036] The final evaluation score is determined by calculating the difference between the first evaluation score and the second evaluation score.

[0037] Secondly, this disclosure also provides a data processing apparatus, comprising:

[0038] The acquisition module is used to acquire the first and second behavior data of at least one recommended user.

[0039] The determination module is used to input the first behavioral data into the positive evaluation model to determine a first evaluation score for positively evaluating the recommendation effect, and to input the second behavioral data into the reverse evaluation model to determine a second evaluation score for reversely evaluating the recommendation effect.

[0040] The evaluation module is used to determine the final evaluation score based on the first evaluation score and the second evaluation score.

[0041] Thirdly, this disclosure also provides an electronic device, including: a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and when the machine-readable instructions are executed by the processor, a data processing method as described in any one of the first aspects and various embodiments thereof is performed.

[0042] Fourthly, this disclosure also provides a computer-readable storage medium storing a computer program that, when executed by a processor, performs the data processing method as described in any one of the first aspects and its various embodiments.

[0043] By employing the aforementioned data processing method, apparatus, electronic device, and storage medium, upon acquiring first and second behavioral data of at least one recommended user, a first evaluation score for positively evaluating the recommendation effect can be determined based on the first behavioral data of at least one recommended user. Conversely, a second evaluation score for negatively evaluating the recommendation effect can be determined based on the second behavioral data of at least one recommended user. Finally, a final evaluation score is determined based on both the first and second evaluation scores. This disclosure combines the first evaluation score for positive evaluation and the second evaluation score for negative evaluation to assess the recommendation effect, resulting in a more comprehensive evaluation and better overall evaluation performance.

[0044] Other advantages of this disclosure will be explained in more detail below with reference to the accompanying drawings.

[0045] It should be understood that the above description is merely an overview of the technical solution of this disclosure, so as to enable a clearer understanding of the technical means of this disclosure, and thus enable its implementation in accordance with the contents of the specification. To make the above and other objects, features, and advantages of this disclosure more apparent and understandable, specific embodiments of this disclosure are illustrated below. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. These drawings are incorporated in and constitute a part of this specification. They illustrate embodiments conforming to this disclosure and, together with the specification, serve to explain the technical solutions of this disclosure. It should be understood that the following drawings only show some embodiments of this disclosure and should not be considered as limiting the scope. Those skilled in the art can obtain other related drawings based on these drawings without creative effort. Furthermore, the same reference numerals denote the same components throughout the drawings. In the drawings:

[0047] Figure 1 A flowchart of a data processing method provided by an embodiment of this disclosure is shown;

[0048] Figure 2 An example diagram is shown illustrating the determination of a first evaluation score in the data processing method provided in this embodiment of the disclosure;

[0049] Figure 3 An example diagram is shown illustrating the determination of direct risk assessment sub-scores in the data processing method provided in this embodiment of the disclosure;

[0050] Figure 4 An example diagram is shown illustrating the determination of indirect risk assessment sub-scores in the data processing method provided in this embodiment of the disclosure;

[0051] Figures 5(a) to 5(b) An example diagram is shown illustrating the updating of evaluation scores in the data processing method provided in this embodiment of the disclosure;

[0052] Figures 6(a) to 6(b) The figure shows a specific example of the data processing method provided in the embodiments of this disclosure;

[0053] Figure 7 A schematic diagram of a data processing apparatus provided in an embodiment of this disclosure is shown;

[0054] Figure 8 A schematic diagram of an electronic device provided in an embodiment of the present disclosure is shown. Detailed Implementation

[0055] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0056] In the description of embodiments disclosed herein, it should be understood that terms such as “comprising” or “having” are intended to indicate the presence of features, numbers, steps, behaviors, components, portions or combinations thereof disclosed herein, and are not intended to exclude the possibility of the presence of one or more other features, numbers, steps, behaviors, components, portions or combinations thereof.

[0057] Unless otherwise stated, " / " means "or". For example, A / B can mean A or B. In this article, "and / or" is merely a way of describing the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone.

[0058] The terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of embodiments of this disclosure, unless otherwise stated, "a plurality of" means two or more.

[0059] Research has revealed that the relevant technologies offer several recommended evaluation schemes.

[0060] One method and apparatus for evaluating the effectiveness of a recommendation model are provided. The method includes: acquiring first historical interaction data and historical behavior logs of each test user for each piece of recommended data, and determining the text sentiment value of each test user for each piece of recommended data based on the first historical interaction data and historical behavior logs; acquiring second historical interaction data and historical relationship data of each test user for the uploader of each piece of recommended data in the test dynamic set, and determining the user sentiment value of each test user for the uploader of each piece of recommended data based on the first historical interaction data, the second historical interaction data, and the historical relationship data; determining the sentiment tendency value of each test user for each piece of recommended data based on the text sentiment value and the user sentiment value; and determining the recommendation effect of the recommendation model based on the sentiment tendency values ​​of a first preset number of test users for a second preset number of pieces of recommended data. This application can improve accuracy.

[0061] Secondly, this invention provides a method, apparatus, system, and computer program product for evaluating recommendation effectiveness. The method includes: acquiring a target user sample set; inputting the target user sample set into a recommendation system to obtain a recommendation result set; determining evaluation influencing factors; and evaluating the recommendation effectiveness of the recommendation system based on the evaluation influencing factors, the recommendation result set, and the target user sample set to obtain a recommendation effectiveness evaluation result. This invention's method incorporates evaluation influencing factors that can affect the evaluation result, avoiding errors caused by these factors in the evaluation process and improving the accuracy of evaluating the recommendation effectiveness of the recommendation system.

[0062] Thirdly, it provides a fusion intelligence analysis method based on credibility and dissemination volume, including: single-event credibility and dissemination volume analysis; multi-event summary type and relative value analysis; and analysis result display. The fusion intelligence analysis method based on credibility and dissemination volume described in this invention has many advantages, such as accurate analysis results, simple steps, and ease of use.

[0063] It is known that current recommendation effectiveness evaluation methods primarily employ time-series classification analysis to assess user activity and retention, lacking qualitative and quantitative evaluation of the correlation between recommended users. Notably, the number of new users referred during the recommendation process reflects both the recommendation's effectiveness and potential risks; however, the lack of current evaluation methods for the number of people involved in the referral process leaves significant room for optimization in related user profile construction and business support.

[0064] To at least partially address one or more of the aforementioned problems and other potential issues, this disclosure provides at least one recommendation evaluation scheme to enrich the current scoring system and form a reasonable value assessment method that balances recommendation effectiveness and risk. This method can reasonably characterize the evolutionary relationship between recommendation effectiveness and risk trends during the fission process of the recommendation propagation network, and can serve as an evaluation factor to support the design and effectiveness monitoring of marketing recommendation strategies.

[0065] To facilitate understanding of this embodiment, a data processing method disclosed in this disclosure will first be described in detail. The data processing method provided in this disclosure is generally executed by an electronic device with a certain computing capability. This electronic device may include, for example, a terminal device, a server, or other processing devices. The terminal device may be a user equipment (UE), a mobile device, a user terminal, a cellular phone, a personal digital assistant (PDA), a wearable device, etc. In some possible implementations, this data processing method can be implemented by a processor calling computer-readable instructions stored in memory.

[0066] See Figure 1 The diagram shows a flowchart of a data processing method provided in an embodiment of this disclosure. The method includes steps S101 to S103, wherein:

[0067] S101: Obtain the first and second behavioral data of at least one recommended user;

[0068] S102: Input the first row of data into the positive evaluation model to determine the first evaluation score for positively evaluating the recommendation effect, and input the second row of data into the negative evaluation model to determine the second evaluation score for negatively evaluating the recommendation effect.

[0069] S103: Determine the final evaluation score based on the first evaluation score and the second evaluation score.

[0070] To facilitate understanding of the data processing method provided in this disclosure, the application scenarios of the method will be described in detail below. The data processing method in this disclosure can be applied to any application scenario requiring recommendation and evaluation. For example, in the application of live-streaming e-commerce, the data processing method provided in this disclosure can be used to evaluate the sales performance of the streamer. Streamers with higher final scores have better sales performance, and vice versa. Here, streamers with better sales performance can be allocated more high-quality resources, thereby optimizing the overall live-streaming e-commerce environment.

[0071] The reason this disclosure employs a two-tiered recommendation evaluation is primarily because recommendation effectiveness is often influenced by various factors. For example, recommending more users may indicate a better recommendation effect to some extent, but in practice, more recommended users may experience various adverse effects, significantly diminishing the overall recommendation effectiveness. To achieve a more comprehensive recommendation evaluation, this disclosure provides a scheme combining two tiers. On one hand, a higher first evaluation score for positively evaluating the recommendation effect indicates a better recommendation effect to some extent; on the other hand, a higher second evaluation score for negatively evaluating the recommendation effect indicates a worse recommendation effect to some extent. In practical applications, the final evaluation score can be determined based on the difference between the first and second evaluation scores.

[0072] The first evaluation score can be determined based on the first behavioral data of at least one recommended user, while the second evaluation score can be determined based on the second behavioral data of at least one recommended user. The first and second behavioral data differ; the former focuses on evaluating the characteristics of an individual recommended user. Taking live-streaming e-commerce as an example, the first behavioral data here can include indicators such as whether the recommended user has sustained purchasing power, user activity, and product attention. To a certain extent, recommended users with higher activity and purchasing power will have higher first evaluation scores. The latter focuses on the correlation characteristics between multiple recommended users. For example, whether there are related transactions between multiple recommended users corresponding to a single recommending user, and the nature of the transaction risks. To a certain extent, recommended users with risky transactions will have higher second evaluation scores.

[0073] In practical applications, the node relationships between recommending users and recommended users can be extracted from relevant recommender networks. For example, the relationship between a single recommending user (corresponding to one recommending node) and multiple recommended users (each recommended user corresponding to one recommended node) can be determined from a typical recommender network where a single user promotes multiple users. Thus, a positive evaluation can be performed based on the first set of data; the higher the first evaluation score, the higher the overall final evaluation score. Conversely, a negative evaluation can be performed based on the second set of data; the higher the second evaluation score, the lower the overall final evaluation score.

[0074] Considering the crucial role of determining the first and second evaluation scores in assessing the effectiveness of the recommendations, the following two aspects will provide a detailed explanation.

[0075] Firstly, the embodiments of this disclosure can determine the first evaluation score according to the following steps:

[0076] Step 1: For each recommended user among at least one recommended user group, obtain the weight coefficient of the indicator corresponding to each positive behavior indicator; and,

[0077] Step 2: Input multiple positive behavior indicators and the corresponding indicator weight coefficients into the positive evaluation model, and determine the first evaluation score for each recommended user by weighted summation;

[0078] Step 3: Sum the first evaluation scores determined for each recommended user to determine the first evaluation score used to positively evaluate the recommendation effect.

[0079] Here, for each recommended user, the weight coefficients of multiple positive behavior indicators can be determined. In this way, the positive evaluation model can be used to sum the multiple positive behavior indicators and their corresponding weight coefficients to determine the first evaluation score for each recommended user. Then, the first evaluation scores of each recommended user are summed to obtain the final first evaluation score.

[0080] To better understand the process of determining the first assessment score, we can then combine it with... Figure 2 Provide an example.

[0081] like Figure 2 As shown, let M be the recommending user node and N be the recommended user node. i The number of recommended users is k. For each recommendation relationship generated, M and N are established. i The associated edge E{MN i}

[0082] Based on the evaluation metrics of the recommended user's recommendation performance, we define the weight values ​​of the associated edges (corresponding to the first evaluation score for a recommendation relationship). Assume the set of relevant metrics is... There are g indicators, and the weight coefficient of each indicator is α. j Then its edge weight value is defined as like Figure 2 As shown, by summing the weights of the k associated edges, the final first evaluation score can be obtained. This score can be used as a positive evaluation coefficient in the evaluation of recommendation effectiveness, expressed as T = ∑V M-Ni This value represents the positive recommendation effect, such as the activity and retention of the recommended users, and it will continue to accumulate and increase as the number of recommended users increases.

[0083] Secondly, the second evaluation score can be determined according to the following steps in the embodiments of this disclosure:

[0084] Step 1: For the first recommended user among multiple recommended users, determine the second behavioral data of the first recommended user and the second behavioral data of the second recommended user associated with the first recommended user;

[0085] Step 2: Input the second behavior data of the first recommended user and the second recommended user into the reverse evaluation model to determine the risk assessment score corresponding to the first recommended user;

[0086] Step 3: Summarize the risk assessment scores of multiple recommended users to determine the second assessment score.

[0087] Here, we can first determine the risk assessment score corresponding to the recommended user, and then summarize the risk assessment scores corresponding to each recommended user to obtain the second assessment score.

[0088] The risk assessment score for the recommended user can be determined based on the second behavioral data of that recommended user and the second behavioral data of other recommended users associated with that user. The first recommended user can be any of the recommended users, all of them, or only a subset of them; there are no specific restrictions here.

[0089] The recommended users mentioned here can be directly associated, such as node users with direct transaction connections, or indirectly associated, such as node users with similar characteristics. In this embodiment, the risk assessment score can be determined by combining the recommended users identified through direct and indirect associations to ensure the accuracy of the assessment results as much as possible. This can be achieved through the following steps:

[0090] Step 1: Based on the risk transaction data of each recommended user, if a direct relationship is found between the first recommended user and the second recommended user, determine the direct risk assessment sub-score corresponding to the first recommended user.

[0091] Step 2: Based on the risk attribute data of each recommended user, determine the indirect risk assessment sub-score between the first recommended user and other recommended users among the multiple recommended users excluding the first recommended user;

[0092] Step 3: Determine the risk assessment score corresponding to the first recommended user based on the direct risk assessment sub-score and the indirect risk assessment sub-score.

[0093] The second set of data here includes risk transaction data and risk attribute data. Based on the risk transaction data, direct risk assessment sub-scores can be determined between users at each node, while indirect risk assessment sub-scores can be determined based on the risk attribute data. The direct risk assessment sub-score is used to evaluate the direct risk level between recommended users, while the indirect risk assessment sub-score is used to evaluate the indirect risk level between recommended users. Regardless of whether it's direct or indirect risk, a higher score generally indicates a poorer overall recommendation effect.

[0094] The following will be the process of determining the two sub-scores (i.e., the direct risk assessment sub-score and the indirect risk assessment sub-score) through the following two aspects.

[0095] Firstly, the embodiments of this disclosure can determine the direct risk assessment sub-score according to the following steps:

[0096] Step 1: If a direct association is found between the first recommended user and at least one second recommended user, determine the direct risk assessment sub-score of the first recommended user relative to each second recommended user;

[0097] Step 2: Sum the direct risk assessment sub-scores relative to at least one second recommended user to determine the direct risk assessment sub-score corresponding to the first recommended user.

[0098] Here, if it is determined that there is a direct association between the first recommended user and one or more second recommended users, the direct risk assessment sub-score of the first recommended user relative to each second recommended user can be determined. Thus, by summing the direct risk assessment sub-scores relative to each second recommended user, the direct risk assessment sub-score corresponding to the first recommended user can be determined.

[0099] In practical applications, there can be one or more direct risk assessment sub-scores. This is mainly because there may be different types of direct risks among the recommended users, and the weights assigned to different types of direct risks may also be different. Therefore, in determining the direct risk assessment sub-score of the first recommended user relative to each second recommended user, we can first obtain multiple edge weight coefficients between the first and second recommended users. Then, the direct risk assessment sub-score of the first recommended user relative to each second recommended user is determined by weighted summation of the multiple direct risk assessment sub-scores and the edge weight coefficients corresponding to each direct risk assessment sub-score.

[0100] In this way, by summing the direct risk assessment sub-scores relative to at least one second recommended user, the direct risk assessment sub-score corresponding to the first recommended user can be determined.

[0101] To better understand the process of determining the sub-scores for the direct risk assessment described above, the following example will illustrate the process:

[0102] like Figure 3 As shown, for a recommended user node N j and N k If high-frequency trading or similar behaviors are observed, a risk correlation edge is established, and a relevant risk weight value of l is assigned, i.e. Different weight values ​​form a set Its edge weight coefficient is defined as β i For N j and N k The associated edges V between nodes jk Weighted summation is defined as follows: Node N can then be obtained. j relative node N kThe direct risk assessment sub-score.

[0103] Secondly, the embodiments of this disclosure can determine the indirect risk assessment sub-score according to the following steps:

[0104] Here, for the other recommended users among multiple recommended users besides the first recommended user, the indirect risk assessment sub-score is determined based on the data similarity between the risk attribute data of the other recommended users and the risk attribute data of the first recommended user.

[0105] To a certain extent, the higher the similarity between data, the higher the corresponding indirect risk assessment sub-score; conversely, the lower the similarity between data, the lower the corresponding indirect risk assessment sub-score.

[0106] To better understand the process of determining the sub-scores for the indirect risk assessment described above, the following example will illustrate the process:

[0107] For different recommended users j and k, their risk performance is measured using risk attribute data. For node N... j and node N k If their behaviors are similar, they pose a potential risk of convergence and spread, according to node N. j Risk attribute data F j and node F k The similarity between risk attribute data defines their risk association similarity, and defines their contagion risk edge. Risk similarity can be calculated by combining Euclidean distance.

[0108] By comparing each recommended user one by one, the risk-similar edges are constructed into a fully connected risk association graph, such as... Figure 4 The dashed line in the middle represents...

[0109] After determining the direct risk assessment sub-score and the indirect risk assessment sub-score, the initial score corresponding to the direct risk assessment sub-score can be updated based on the indirect risk assessment sub-score to determine the risk assessment score corresponding to the first recommended user.

[0110] During the score update process, we can first determine the product of the indirect risk assessment sub-scores corresponding to other recommended users and the initial scores of other recommended users. A larger product indicates a higher probability of indirect risk for other recommended users corresponding to the first recommended user; conversely, a smaller product indicates a lower probability of indirect risk for other recommended users corresponding to the first recommended user. Based on this, we can select the maximum product value to update the initial score, thereby determining the risk assessment score corresponding to the first recommended user.

[0111] To better understand the above score update process, we can combine it with... Figures 5(a) to 5(b) And an example to illustrate this point.

[0112] For each risk node j, as shown in Figure 5(a), the initial risk value of risk node j is its own direct risk index value F. j The sum of the risk transaction edge weights associated with j and having factual risk, i.e.:

[0113]

[0114] The initial risk value of each node j can be calculated from this:

[0115] Here, for each recommended user j, the risk similarity of other nodes in the network and the risk contagion similarity boundary value are considered. Update its risk value FR, as shown in Figure 5(b), which is the sum of the current node's risk and the risk of its largest neighbor node weighted by similarity:

[0116]

[0117] This round of propagation can update the potential risk propagation risk of each node (corresponding to the indirect risk assessment sub-score).

[0118] Here, by summing the updated risk values ​​at each node, the potential propagation risk (corresponding to the second evaluation score) for each recommended user is obtained:

[0119] Fr=∑Fr' j0

[0120] This round of calculations shows that as the number of node users increases, the potential risk of transmission also increases when the risk of related users is highly similar.

[0121] Based on the positive evaluation coefficient T and the negative evaluation coefficient Fr, the following final evaluation score can be calculated: γ=(T-Fr) / x, where x corresponds to the number of recommended users, which can better describe the risk propagation recommendation effect and risk impact of the recommendation network.

[0122] To facilitate a further understanding of the data processing method provided in the embodiments of this disclosure, the following example uses UnionPay QuickPass as a recommended scenario for illustration.

[0123] For the scenario of recommending 6 people, the following recommendation scheme can be designed: Figures 6(a) to 6(b) As shown.

[0124] Assume that each recommended user has a boundary value of 0.7, as shown in Figure 6(a). The behavioral similarity between recommended users is 0.8. Some users have risk association relationships with a boundary value of 0.3. The specific score update process is shown in Figure 6(b).

[0125] With 5 recommended users, k = 6. Here, the first evaluation score for positive evaluation is T = 5 * 0.7 = 3.5, and the second evaluation score for negative evaluation is Fr = 3.336; thus, the final evaluation score is γ = (T - Fr) / 6 = 0.027.

[0126] As can be seen, the data processing method provided by the embodiments of this disclosure can effectively reflect the increased risk of contagion caused by the increase in the number of users, and the corresponding recommendation evaluation factor γ value also reflects a large downward trend, resulting in better evaluation performance.

[0127] In the description of this specification, the references to terms such as "some possible implementations," "some implementations," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that implementation or example is included in at least one implementation or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same implementation or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more implementations or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different implementations or examples described in this specification, as well as the features of different implementations or examples.

[0128] Regarding the method flowcharts of embodiments of this disclosure, certain operations are described as different steps performed in a certain order. Such flowcharts are illustrative and not restrictive. Some steps described herein may be grouped together and performed in a single operation, some steps may be divided into multiple sub-steps, and some steps may be performed in an order different from that shown herein. The various steps shown in the flowcharts may be implemented in any way by any circuit structure and / or tangible mechanism (e.g., software running on a computer device, hardware (e.g., logic functions implemented by a processor or chip), and / or any combination thereof).

[0129] Those skilled in the art will understand that, in the above-described method of the specific implementation, the order in which each step is written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0130] Based on the same inventive concept, this disclosure also provides a data processing device corresponding to the data processing method. Since the principle of the device in this disclosure for solving the problem is similar to that of the data processing method described above, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0131] Reference Figure 7 The diagram shown is a schematic representation of a data processing apparatus provided in an embodiment of this disclosure. The apparatus includes: an acquisition module 701, a determination module 702, and an evaluation module 703; wherein,

[0132] The acquisition module 701 is used to acquire the first behavior data and the second behavior data of at least one recommended user;

[0133] The determination module 702 is used to input the first row of data into the positive evaluation model to determine the first evaluation score for positively evaluating the recommendation effect, and to input the second row of data into the reverse evaluation model to determine the second evaluation score for reversely evaluating the recommendation effect.

[0134] The evaluation module 703 is used to determine the final evaluation score based on the first evaluation score and the second evaluation score.

[0135] Using the aforementioned data processing apparatus, upon acquiring first and second behavioral data of at least one recommended user, a first evaluation score for positively evaluating the recommendation effect can be determined based on the first behavioral data of at least one recommended user. Conversely, a second evaluation score for negatively evaluating the recommendation effect can be determined based on the second behavioral data of at least one recommended user. Then, based on the first and second evaluation scores, a final evaluation score is determined. This disclosure combines the first evaluation score for positive evaluation and the second evaluation score for negative evaluation to assess the recommendation effect, resulting in a more comprehensive evaluation and better overall evaluation performance.

[0136] In one possible implementation, when there are multiple recommended users, the determining module 702 is configured to determine a second evaluation score for reverse evaluation of the recommendation effect according to the following steps:

[0137] For the first recommended user among multiple recommended users, determine the second behavioral data of the first recommended user and the second behavioral data of the second recommended user associated with the first recommended user;

[0138] Input the second behavioral data of the first recommended user and the second behavioral data of the second recommended user into the reverse evaluation model to determine the risk assessment score corresponding to the first recommended user;

[0139] The risk assessment scores of multiple recommended users are aggregated to determine the second assessment score.

[0140] In one possible implementation, when the second behavioral data includes risk transaction data and risk attribute data, the determining module 702 is used to determine the risk assessment score corresponding to the first recommended user according to the following steps:

[0141] If a direct correlation is found between the first recommended user and the second recommended user based on the risk transaction data of each recommended user, then the direct risk assessment sub-score corresponding to the first recommended user is determined.

[0142] Based on the risk attribute data of each recommended user, determine the indirect risk assessment sub-score between the first recommended user and other recommended users among the multiple recommended users besides the first recommended user;

[0143] The risk assessment score for the first recommended user is determined based on the direct risk assessment sub-score and the indirect risk assessment sub-score.

[0144] In one possible implementation, the determining module 702 is configured to determine the direct risk assessment sub-score corresponding to the first recommended user according to the following steps:

[0145] If a direct association is found between the first recommended user and at least one second recommended user, the direct risk assessment sub-score of the first recommended user relative to each second recommended user is determined.

[0146] The direct risk assessment sub-score corresponding to the first recommended user is determined by summing the direct risk assessment sub-scores relative to at least one second recommended user.

[0147] In one possible implementation, where each second recommended user includes multiple direct risk assessment sub-scores, the determining module 702 is configured to determine the direct risk assessment sub-score of the first recommended user relative to each second recommended user according to the following steps:

[0148] For each second recommended user, obtain multiple edge weight coefficients between the first recommended user and the second recommended user;

[0149] The direct risk assessment sub-score of the first recommended user relative to the second recommended user is determined by a weighted sum of multiple direct risk assessment sub-scores relative to the second recommended user and the edge weight coefficients corresponding to each direct risk assessment sub-score.

[0150] In one possible implementation, the determining module 702 is configured to determine the indirect risk assessment sub-score between the first recommended user and other recommended users among a plurality of recommended users, according to the following steps:

[0151] For the other recommended users among multiple recommended users besides the first recommended user, the indirect risk assessment sub-score is determined based on the data similarity between the risk attribute data of the other recommended users and the risk attribute data of the first recommended user.

[0152] In one possible implementation, the determining module 702 is configured to determine the risk assessment score corresponding to the first recommended user based on the direct risk assessment sub-score and the indirect risk assessment sub-score corresponding to the first recommended user, according to the following steps:

[0153] The initial score of the first recommended user is determined based on the direct risk assessment sub-score corresponding to the first recommended user and the first assessment score of the first recommended user.

[0154] The initial score is updated based on the indirect risk assessment sub-score to determine the risk assessment score corresponding to the first recommended user.

[0155] In one possible implementation, when there are multiple other recommended users, the determining module 702 is used to update the initial score based on the indirect risk assessment sub-score according to the following steps to determine the risk assessment score corresponding to the first recommended user:

[0156] For each of the multiple other recommended users, the indirect risk assessment sub-score corresponding to the other recommended user is multiplied by the initial score of the other recommended user to determine the product result;

[0157] The maximum value of the product is selected to update the initial score, thus determining the risk assessment score corresponding to the first recommended user.

[0158] In one possible implementation, when the first behavioral data includes multiple positive behavioral indicators, the determining module 702 is configured to determine a first evaluation score for positively evaluating the recommendation effect according to the following steps:

[0159] For each of the at least one recommended users, obtain the weight coefficient of the indicator corresponding to each positive behavior indicator; and,

[0160] Multiple positive behavior indicators and their corresponding weight coefficients are input into the positive evaluation model, and the first evaluation score for each recommended user is determined by weighted summation.

[0161] The first evaluation score determined for each recommended user will be summed to determine the first evaluation score used to positively evaluate the recommendation effect.

[0162] In one possible implementation, the evaluation module 703 is configured to determine the final evaluation score based on the first evaluation score and the second evaluation score according to the following steps:

[0163] The final evaluation score is determined by calculating the difference between the first evaluation score and the second evaluation score.

[0164] It should be noted that the apparatus in the embodiments of this application can implement each process of the aforementioned method and achieve the same effect and function, which will not be elaborated here.

[0165] This disclosure also provides an electronic device, such as... Figure 8 The diagram shown is a schematic representation of an electronic device structure provided in this embodiment of the present disclosure, including: a processor 801, a memory 802, and a bus 803. The memory 802 stores machine-readable instructions executable by the processor 801 (e.g., ...). Figure 7 The device includes modules 701 for acquiring, 702 for determining, and 703 for evaluating the corresponding execution instructions. When the electronic device is running, the processor 801 communicates with the memory 802 via bus 803. When a machine-readable instruction is executed by the processor 801, the following processing is performed:

[0166] Obtain the first and second behavioral data of at least one recommended user;

[0167] The first row of data is input into the positive evaluation model to determine the first evaluation score used to positively evaluate the recommendation effect, and the second row of data is input into the negative evaluation model to determine the second evaluation score used to negatively evaluate the recommendation effect.

[0168] The final evaluation score is determined based on the first evaluation score and the second evaluation score.

[0169] This disclosure also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the data processing method described in the above method embodiments. The storage medium may be a volatile or non-volatile computer-readable storage medium.

[0170] This disclosure also provides a computer program product carrying program code. The program code includes instructions that can be used to execute the steps of the data processing method described in the above method embodiments. For details, please refer to the above method embodiments, which will not be repeated here.

[0171] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium; in another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0172] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. In particular, the descriptions of the apparatus, device, and computer-readable storage medium embodiments are simplified because they are substantially similar to the method embodiments; relevant details can be found in the descriptions of the method embodiments.

[0173] The apparatus, device, and computer-readable storage medium provided in this application correspond one-to-one with the method. Therefore, the apparatus, device, and computer-readable storage medium also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the apparatus, device, and computer-readable storage medium will not be repeated here.

[0174] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, apparatus (devices or systems), or computer-readable storage media. Therefore, this disclosure can take the form of a completely hardware implementation, a completely software implementation, or an implementation combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer-readable storage medium implemented on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-readable program code.

[0175] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices or systems), and computer-readable storage media according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0176] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0177] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0178] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0179] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0180] Computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device. Furthermore, although the operations of the methods of this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all of the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0181] While the spirit and principles of this disclosure have been described with reference to several specific embodiments, it should be understood that this disclosure is not limited to the disclosed specific embodiments, and the division of aspects does not imply that features in these aspects cannot be combined for benefit; such division is merely for convenience of expression. This disclosure is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims

1. A data processing method, characterized by, The method comprises: obtaining first behavior data and second behavior data of at least one recommended user, wherein the first behavior data comprises a plurality of positive behavior indicators for positively evaluating the recommendation effect, and the second behavior data comprises risk transaction data and risk attribute data for negatively evaluating the recommendation effect; inputting the first behavior data into a positive evaluation model, performing weighted summation on the plurality of positive behavior indicators by the positive evaluation model to determine a first evaluation score for positively evaluating the recommendation effect, and inputting the second behavior data into a negative evaluation model to determine a second evaluation score for negatively evaluating the recommendation effect based on the risk transaction data and the risk attribute data by the negative evaluation model; determining a final evaluation score based on the first evaluation score and the second evaluation score.

2. The method of claim 1, wherein, In the case of multiple recommended users, the determination of the second evaluation score for negatively evaluating the recommendation effect comprises: for a first recommended user among the multiple recommended users, determining second behavior data of the first recommended user and second behavior data of a second recommended user associated with the first recommended user; inputting the second behavior data of the first recommended user and the second behavior data of the second recommended user into the negative evaluation model to determine a risk evaluation score corresponding to the first recommended user; summarizing the risk evaluation scores corresponding to the multiple recommended users respectively to determine the second evaluation score.

3. The method of claim 2, wherein, The determination of the risk evaluation score corresponding to the first recommended user comprises: in the case of determining that there is a direct association between the first recommended user and the second recommended user based on the risk transaction data of each recommended user, determining a direct risk evaluation sub-score corresponding to the first recommended user; based on the risk attribute data of each recommended user, determining an indirect risk evaluation sub-score between the first recommended user and other recommended users except the first recommended user among the multiple recommended users; based on the direct risk evaluation sub-score and the indirect risk evaluation sub-score corresponding to the first recommended user, determining the risk evaluation score corresponding to the first recommended user.

4. The method of claim 3, wherein, The determination of the direct risk evaluation sub-score corresponding to the first recommended user comprises: in the case of determining that there is a direct association between the first recommended user and at least one second recommended user, determining a direct risk evaluation sub-score of the first recommended user relative to each second recommended user; summing the direct risk evaluation sub-scores relative to at least one second recommended user to determine the direct risk evaluation sub-score corresponding to the first recommended user.

5. The method of claim 4, wherein, In the case of including a plurality of direct risk evaluation sub-scores relative to each second recommended user, the determination of the direct risk evaluation sub-score of the first recommended user relative to each second recommended user comprises: for each second recommended user, obtaining a plurality of edge weight coefficients between the first recommended user and the second recommended user; determine a direct risk assessment sub-score of the first recommended user relative to the second recommended user based on a weighted sum between a plurality of direct risk assessment sub-scores relative to the second recommended user and an edge weight coefficient corresponding to each direct risk assessment sub-score.

6. The method of claim 3, wherein, The determining of the indirect risk assessment sub-score between the first recommended user and other recommended users in the plurality of recommended users except the first recommended user comprises: For other recommended users in the plurality of recommended users except the first recommended user, determining the indirect risk assessment sub-score based on a data similarity between risk attribute data of the other recommended user and risk attribute data of the first recommended user.

7. The method according to any one of claims 3 to 6, characterized in that, The determining of the risk assessment score corresponding to the first recommended user based on the direct risk assessment sub-score and the indirect risk assessment sub-score corresponding to the first recommended user comprises: determining an initial score of the first recommended user based on the direct risk assessment sub-score corresponding to the first recommended user and the first assessment score of the first recommended user; updating the initial score according to the indirect risk assessment sub-score to determine the risk assessment score corresponding to the first recommended user.

8. The method of claim 7, wherein, In the case that the other recommended user is a plurality of other recommended users, the updating of the initial score according to the indirect risk assessment sub-score to determine the risk assessment score corresponding to the first recommended user comprises: For each of the plurality of other recommended users, multiplying the indirect risk assessment sub-score corresponding to the other recommended user and the initial score of the other recommended user to determine a product result; selecting a maximum value of the product result to update the initial score to determine the risk assessment score corresponding to the first recommended user.

9. The method of claim 1, wherein, The determining of the first assessment score for positively evaluating the recommendation effect comprises: For each of the at least one recommended user, obtaining an index weight coefficient corresponding to each of the positive behavior indicators; and inputting the plurality of positive behavior indicators and the index weight coefficient corresponding to each positive behavior indicator into a positive evaluation model to determine the first assessment score for each of the recommended user by weighted sum; summing the first assessment score determined for each of the recommended user to determine the first assessment score for positively evaluating the recommendation effect.

10. The method of claim 1, wherein, The determining of the final assessment score based on the first assessment score and the second assessment score comprises: determining the final assessment score by difference between the first assessment score and the second assessment score.

11. A data processing apparatus, characterized by comprise: an acquisition module configured to acquire first behavior data and second behavior data of at least one recommended user, wherein the first behavior data comprises a plurality of positive behavior indicators for positively evaluating a recommendation effect, and the second behavior data comprises risk transaction data and risk attribute data for negatively evaluating the recommendation effect; The determining module is configured to input the first behavior data into a forward evaluation model, perform weighted summation on the plurality of forward behavior indexes through the forward evaluation model, determine a first evaluation score for forward evaluation of the recommended effect, and input the second behavior data into a reverse evaluation model, determine a second evaluation score for reverse evaluation of the recommended effect based on the risk transaction data and risk attribute data through the reverse evaluation model; The evaluation module is configured to determine a final evaluation score based on the first evaluation score and the second evaluation score.

12. An electronic device, comprising: The data processing method comprises the following steps: A processor, a memory and a bus, the memory stores machine readable instructions executable by the processor, when the electronic device is running, the processor and the memory communicate through the bus, the machine readable instructions are executed by the processor to execute the data processing method in any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to execute the data processing method in any one of claims 1 to 10.

Images