Data processing method and device, electronic equipment and storage medium

Abstract

The application discloses a data processing method and device, electronic equipment and storage medium. The application relates to the technical field of artificial intelligence. The method comprises the following steps: obtaining historical behavior data and user feature data of a target user; processing the user feature data based on an integrated algorithm to obtain a first demand degree of the target user for a target product; performing collaborative filtering processing on the historical behavior data to determine a second demand degree of the target user for the target product; determining a third demand degree of the target user for the target product according to the user feature data and at least one pre-set evaluation dimension; and determining a target demand degree of the target user for the target product based on the first demand degree, the second demand degree, the third demand degree and corresponding weight values. The application solves the problem that the products with high transaction frequencies are recommended to users in the prior art, leading to the inadaptability of the recommended products to the users, improves the adaptability between the recommended target products and the target users, and meets the user demand.

Description

Technical Field

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

[0002] Currently, potential products that match the user's needs are often recommended to meet user requirements and improve user experience. The current product recommendation method generally uses feature data extracted from the user's historical transaction information to recommend products with high transaction frequency to the user, or recommend other products with high similarity to products with high transaction frequency to the user.

[0003] The problem with this method is that it can only frequently recommend products that the user already owns, and it cannot accurately predict each user's preferences, resulting in low accuracy of the recommendation results and the problem that the recommended products often do not match the user's actual preferences. Summary of the Invention

[0004] This invention provides a data processing method, apparatus, electronic device, and storage medium to improve the accuracy of target demand determination while ensuring that the target product determined based on the target demand is matched with the target user, thereby achieving the technical effect of meeting user needs.

[0005] In a first aspect, embodiments of the present invention provide a data processing method, the method comprising:

[0006] Acquire historical behavioral data and user characteristic data of the target users;

[0007] The user feature data is processed based on the ensemble algorithm to obtain the target user's first degree of demand for the target product;

[0008] The historical behavior data is subjected to collaborative filtering to determine the target user's second degree of demand for the target product.

[0009] Based on the user characteristic data and at least one pre-set evaluation dimension, the third degree of demand for the target product by the target user is determined;

[0010] Based on the first demand level, the second demand level, the third demand level, and the corresponding weight values, the target user's target demand level for the target product is determined.

[0011] Secondly, embodiments of the present invention also provide a data processing apparatus, the apparatus comprising:

[0012] The data acquisition module is used to acquire historical behavioral data and user characteristic data of the target user;

[0013] The first demand degree determining module is configured to process the user characteristic data based on an integration algorithm to obtain a first demand degree of the target user for the target product;

[0014] The second demand degree determining module is configured to perform collaborative filtering processing on the historical behavior data to determine a second demand degree of the target user for the target product;

[0015] The third demand degree determining module is configured to determine a third demand degree of the target user for the target product according to the user characteristic data and at least one evaluation dimension set in advance;

[0016] The target demand degree determining module is configured to determine a target demand degree of the target user for the target product based on the first demand degree, the second demand degree, the third demand degree and corresponding weight values.

[0017] In a third aspect, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the data processing method according to any of the embodiments of the present application when executing the program.

[0018] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program executable by a processor to implement the data processing method according to any of the embodiments of the present application.

[0019] In a fifth aspect, a computer program product is provided, which includes a computer program executable by a processor to implement the data processing method according to any of the embodiments of the present application.

[0020] The technical scheme of the embodiment of the present application is: historical behavior data and user feature data of a target user are acquired; the user feature data is processed based on an integrated algorithm to obtain a first demand degree of the target user for a target product; the historical behavior data is processed by collaborative filtering to determine a second demand degree of the target user for the target product; the user feature data and at least one pre-set evaluation dimension are used to determine a third demand degree of the target user for the target product; and the first demand degree, the second demand degree, the third demand degree and corresponding weight values are used to determine a target demand degree of the target user for the target product. The present application solves the problem in the prior art that a product with high transaction frequency in user historical transaction information is recommended to a user, resulting in poor recommendation effect and causing the recommended product to be incompatible with the user. The demand degrees determined by the three evaluation methods of integrated algorithm, collaborative filtering and evaluation dimension are integrated, and based on the demand degrees and corresponding weight values, a target demand degree that can effectively and accurately represent the demand degree of the target user for the target product is determined. The target product to be pushed to the target user is determined based on the target demand degree, so that the target product determined to be pushed to the target user is compatible with the target user, the recommendation effect is improved, and the technical effect of meeting the user demand is achieved.

[0021] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0023] Figure 1 is a flowchart of a data processing method according to the first embodiment of the present application;

[0024] Figure 2 is a flowchart of a data processing method according to the second embodiment of the present application;

[0025] Figure 3 is a flowchart of a data processing method according to the third embodiment of the present application;

[0026] Figure 4 is a flowchart of a data processing method according to the fourth embodiment of the present application;

[0027] Figure 5 is a structural schematic diagram of a data processing device according to the sixth embodiment of the present application;

[0028] Figure 6 This is a schematic diagram of the structure of an electronic device that implements the data processing method of the present invention. Detailed Implementation

[0029] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0030] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance. The acquisition, storage, use, and processing of data in the technical solutions of this application all comply with the relevant provisions of national laws and regulations.

[0031] Example 1

[0032] Figure 1 This is a flowchart of a data processing method according to Embodiment 1 of the present invention. This embodiment is applicable to recommended situations. The method can be executed by a data processing device, which can be implemented in hardware and / or software and can be configured in a computing device. Figure 1 As shown, the method includes:

[0033] S110. Obtain historical behavioral data and user characteristic data of the target user.

[0034] In this context, the target user can be understood as a user who needs product recommendations. Historical behavior data may include the relationship between users and products. Optionally, historical behavior data may include historical access records for at least one product to be selected, and at least one product to be selected may include the target product. For example, historical behavior data may include records of users' historical product searches, product clicks, product additions to cart, and product inquiries, as well as information on various product types, prices, sales volume, rankings, and risk records. User characteristic data includes basic user data and asset information. For example, basic user data may consist of fundamental information associated with the user. This fundamental information is specific to each user, meaning that different users may have different fundamental information.

[0035] In the embodiment, the basic information and asset information of the target user can be acquired under the premise that the user privacy is guaranteed, as the user feature data, and the search product data, click product data, add-to-cart product data and consultation product data input by the target user in the past are acquired as the historical behavior data of the target user. Thus, the product suitable for the target user can be recommended by analyzing the historical behavior data and the user feature data of the target user, so as to meet the use demand of the user.

[0036] In S120, the user feature data is processed based on the integrated algorithm to obtain the first demand degree of the target product of the target user.

[0037] The integrated algorithm can be understood as an algorithm integrating multiple classification algorithms, and each classification algorithm can be understood as a classifier. The multiple classification algorithms can be the same or different. For example, the integrated algorithm can be Random forest, Adaboost, lightGBM or XGBoost, or an algorithm integrating at least two of Random forest, Adaboost, lightGBM and XGBoost. It should be noted that the integrated algorithm can be determined by the technical personnel according to the actual work. For example, when the computing resource is insufficient, a relatively simple model such as Random forest can be used, and when the computing resource is sufficient, lightGBM and XgBoost algorithms are recommended, which can improve the operation speed and greatly save the calculation time without reducing the accuracy. Random forest, lightGBM and XgBoost algorithms can also be integrated.

[0038] In actual application, in the process of determining the demand degree of the target product of the target user in the to-be-selected product, the user feature data can be processed by each classification algorithm in the integrated algorithm in combination with the basic information of the target product, so that each classification algorithm predicts the demand degree of the target product of the target user, and multiple probability values are obtained. The first demand degree of the target product of the target user can be obtained by counting the probability values, for example, the average value or mean square error of the probability values can be used as the first demand degree.

[0039] It should be noted that in step S120, each target product in the to-be-selected products will obtain a first demand degree. Assuming that there are target products A, B and C, the product features of product A and the user features of the target user are analyzed by using the integration algorithm to obtain the first demand degree of product A, the product features of product B and the user features of the target user are analyzed to obtain the first demand degree of product B, and the product features of product C and the user features of the target user are analyzed to obtain the first demand degree of product C. Correspondingly, the first demand degrees of the target user for each target product can be obtained.

[0040] S130, performing collaborative filtering processing on the historical behavior data to determine a second demand degree of the target user for the target product.

[0041] In order to improve the effectiveness of the recommendation, the interest preferences of the target user can be discovered according to the historical behavior data of the target user, and products similar to the interest preferences can be recommended to the target user based on the discovered interest preferences of the target user.

[0042] In actual application, the second demand degree of the target user for each target product can be determined by performing collaborative filtering processing on the historical behavior data of the target user. For example, the behavior features in the historical behavior data of the target user can be extracted, other users having similar interests (such as having similar product acquisition records, working in the same industry or having similar annual incomes) to the target user can be found out, the interest similarity between the target user and the other users can be calculated, and accordingly, the demand degree of the target user for the products related to the similar users having higher interest similarity can be higher, and the demand degree of the target user for the products related to the similar users having lower interest similarity can be lower. Other products having similar features (such as similar product categories) to the product related to the target user can also be found out by using the historical behavior data, the similarity between the products can be calculated, and for the products having higher similarity, the demand degree of the target user for the product can be higher, and for the products having lower similarity, the demand degree of the target user for the product can be lower, which is the second demand degree. It should be noted that the behavior feature data of the other users is also obtained under the premise that the other users authorize and the privacy of the other users is ensured.

[0043] S140, determining a third demand degree of the target user for the target product according to the user feature data and at least one evaluation dimension pre-set.

[0044] The evaluation dimension can be a dimension related to the user, such as education level information, income level information and work type information, etc. The evaluation dimension can also be a dimension related to the product, and the specific evaluation dimension can be determined by a technician according to the actual working condition, which is not limited in the technical solution.

[0045] Specifically, the target user can be evaluated in each evaluation dimension according to the basic information and the user feature data corresponding to the target product, to obtain corresponding evaluation values, and then the evaluation values are weighted and fused to obtain a final evaluation value of the target user for the target product as the third demand degree.

[0046] It should be noted that the above S120 to S140 can be executed sequentially or in parallel, and the specific execution order is not limited. The above order is only the order of explaining the technical solutions in each step, not the execution order of each step.

[0047] S150, determining a target demand degree of the target user for the target product based on the first demand degree, the second demand degree, the third demand degree, and the corresponding weight values.

[0048] The target demand degree can be used to represent the demand degree or the preference degree for the product.

[0049] It should be noted that in order to improve the accuracy of the target demand degree, different weight values can be set for different ways of determining the demand degree for the target product, so that after the demand degrees determined based on different ways are weighted and fused, the target demand degree that can accurately represent the demand degree of the target user for the target product is obtained.

[0050] Specifically, when determining the demand degree of the target user for a target product, a first intermediate value can be determined according to the product of the first demand degree and the corresponding weight value; a second intermediate value can be determined based on the product of the second demand degree and the corresponding weight value; a third intermediate value can be determined according to the product of the third demand degree and the corresponding weight value; and the target demand degree can be determined based on the first intermediate value, the second intermediate value, and the third intermediate value.

[0051] In this embodiment, the first demand degree and the corresponding weight value can be multiplied to obtain a product value as the first intermediate value; the second demand degree and the corresponding weight value can be multiplied to obtain a second intermediate value; and the third demand degree and the corresponding weight value can be multiplied to obtain a third intermediate value. Further, the mean or mean square deviation of the first intermediate value, the second intermediate value, and the third intermediate value can be used as the target demand degree. For example, the first demand degree is p1, the corresponding weight value is w1, the second demand degree is p2, the corresponding weight value is w2, and the third demand degree is p3, the corresponding weight value is w3, and the target demand degree s = w1*p1+w2*p2+w3*p3.

[0052] It should be noted that after the target demand degree of the target user for each target product is determined, the product meeting the user demand and preference can be recommended to the user, for example, the target demand degree of the target user for each target product is recommended and sorted in descending order, and each target product can be pushed to the target user in descending order. The pushing mode can also be: if the target demand degree is greater than the preset demand degree threshold, the target product is pushed to the target user; if the target demand degree is less than the preset demand degree threshold, the target product is not pushed to the target user.

[0053] Specifically, the target demand degree can be compared with the preset demand degree threshold, if the target demand degree is greater than the preset demand degree threshold, it is considered that the demand degree and the preference degree of the target user for the target product are high, at this time the target product can be pushed to the target user. If the target demand degree is less than the preset demand degree threshold, it is considered that the demand degree and the preference degree of the target user for the target product are low, at this time the target product is not pushed to the target user. Exemplarily, a demand degree threshold T can be set, when s≥T, it is determined that the demand degree of the user for the target product is high, that is, it can be recommended, when s<T, it is determined that the demand degree of the user for the target product is low, and no recommendation is made.

[0054] The technical scheme of the embodiment, by acquiring the historical behavior data and the user feature data of the target user, processing the user feature data based on the integrated algorithm to obtain the first demand degree of the target user for the target product, performing collaborative filtering processing on the historical behavior data to determine the second demand degree of the target user for the target product, determining the third demand degree of the target user for the target product according to the user feature data and the pre-set at least one evaluation dimension, determining the target demand degree of the target user for the target product based on the first demand degree, the second demand degree, the third demand degree and the corresponding weight value, solves the problem that the products with high transaction frequency in the user historical transaction information are recommended to the user in the prior art, resulting in poor recommendation effect and causing the recommended products to be not adapted to the user, realizes the demand degree determined by the three evaluation methods of comprehensive integrated algorithm, collaborative filtering and evaluation dimension, and based on each demand degree and the corresponding weight value, determines the target demand degree which can effectively and accurately represent the demand degree of the target user for the target product, determines the target product which needs to be pushed to the target user based on the target demand degree, so that the determined target product pushed to the target user is matched with the target user, improves the recommendation effect, and achieves the technical effect of meeting the user demand.

[0055] Embodiment two

[0056] Figure 2is a flow chart of a data processing method according to the second embodiment of the present application. Based on the foregoing embodiment, after the historical behavior data and the user feature data of the target user are obtained, feature extraction can be performed on the historical behavior data and the user feature data respectively to obtain a historical behavior matrix and a user feature matrix, so that the processed data can adapt to the model and improve the data processing efficiency. The specific implementation can be referred to the technical solution of the present embodiment. The same or corresponding technical terms as the foregoing embodiments are not repeated here.

[0057] As shown in Figure 2 , the method specifically comprises the following steps:

[0058] S210, obtaining historical behavior data and user feature data of a target user.

[0059] S220, performing feature extraction on the historical behavior data and the user feature data respectively to obtain a historical behavior matrix corresponding to the historical behavior data and a user feature matrix corresponding to the user feature data.

[0060] The historical behavior matrix can include the relationship between the target user and each product to be selected. The user feature matrix can include the relationship between the target user and each behavior feature.

[0061] In actual application, feature extraction algorithms can be used to extract feature information in the historical behavior data and the user feature data respectively to obtain a historical behavior matrix corresponding to the historical behavior data. For example, the rows in the historical behavior matrix can correspond to the target user, and each column corresponds to product information to be selected. If there is a transaction behavior between user i and product j, the element of the ith row and the jth column is set to 1, otherwise it is 0. For example, the historical behavior matrix is [0 1 1], where the first 0 indicates that the target user has transaction information with product A, the second 1 indicates that the target user has transaction information with product B, and the third 1 indicates that the target user has transaction information with product C. A user feature matrix corresponding to the user feature data can also be obtained. For example, the rows of the user feature matrix correspond to the target user, and each column corresponds to a user behavior feature. The element of the ith (user i) row and the jth (behavior feature j) column can be the attribute information of the behavior feature j. This enables subsequent data processing of the historical behavior matrix and the user feature matrix based on the model to obtain the corresponding demand degree and improve the data processing efficiency.

[0062] It should be noted that, in order to further improve the data processing efficiency, the historical behavior data and the user feature data can be cleaned respectively when the feature extraction is performed, such as deleting or correcting abnormal values, error data, and supplementing missing values. For example, after determining the abnormal values, error data or missing values, the cleaning method of the abnormal values, error data or missing values can be as follows: taking the error data as an example, the Monte Carlo algorithm can be used to analyze the data distribution characteristics based on other normal information associated with the error data, find the distribution function, and then generate new values as the supplementary correction information corresponding to the error data by the distribution function. For example, if a user lacks work information, the work information of the user can be simulated based on the work information of other users similar to the basic information of the user. It is assumed that the work information of other users is found to be approximately normally distributed through distribution histogram and sample experience distribution function analysis, and then a normal distribution random number generator can be used to simulate the work information of the user. If the data distribution of other users is too complex to be simulated by the distribution function, a Markov Chain Monte Carlo (MCMC) algorithm can be used to simulate the correction estimate value.

[0063] S230, inputting the user feature matrix corresponding to the user feature data into the prediction model corresponding to the integrated algorithm to obtain a first demand degree of the target product of the target user.

[0064] The prediction model can be obtained by training a training sample matrix based on the integrated algorithm. The training sample matrix corresponds to the user feature matrix of each user to be trained. For example, the user feature data of a plurality of users to be trained can be obtained under the condition that the users are informed and authorized. The user feature of each user and the product feature of a product corresponding to the user can be combined into a training sample, and the transaction behavior (such as existing transaction behavior or non-existing transaction behavior) between the user and the product information is recorded as the training result of the training sample. A training sample matrix in the same format as the user feature matrix and containing a plurality of training samples and corresponding training results is obtained. The training sample matrix is input into the to-be-trained model to fit the to-be-trained model, fix the model parameters, determine the model structure, and obtain the preset model. It should be noted that, in order to improve the accuracy of the preset model, as many training samples as possible can be obtained.

[0065] In practical applications, the user feature matrix can be taken as an input of the prediction model, and an output prediction result can be a numerical value in the range of 0-1, and the larger the numerical value is, the higher the probability of predicting the user's demand for the target product is. The prediction result can be taken as the first demand degree. It should be noted that if the prediction model includes multiple classification models such as Randomforest, lightGBM, and XGBoost, the prediction probabilities generated by each classification model can be weighted and averaged to obtain the prediction probability of the comprehensive output of the prediction model, which is taken as the first demand degree.

[0066] In S240, a historical behavior matrix corresponding to the historical behavior data is input into a collaborative filtering model to obtain a second demand degree of the target user for the target product.

[0067] The collaborative filtering model is determined based on a predetermined training set data matrix corresponding to the historical behavior matrix of each user to be trained. For example, the historical behavior data of a plurality of users to be trained can be obtained under the authorization of the users, and the corresponding historical behavior matrix can be constructed. The user information of each user and the product information of a corresponding product can be combined into a training sample, and the transaction behavior (such as existing transaction behavior or non-existing transaction behavior) between the user and the product information is recorded as the training result of the training sample. A training set data matrix containing a plurality of training samples and corresponding training results is obtained, and then the training set data matrix is input into the collaborative filtering model to be trained to fit the collaborative filtering model to be trained. After fitting is completed, the trained collaborative filtering model is obtained. It should be noted that in order to improve the accuracy of the collaborative filtering model, as many training set data as possible can be obtained so that the output result of the collaborative filtering model can effectively reflect the demand degree of each user for each product.

[0068] In practical applications, the historical behavior matrix corresponding to the target user can be taken as an input of the collaborative filtering model, and the data in the input historical behavior matrix and the historical behavior matrix of each user to be trained are calculated for similarity, such as calculating the similarity between different users or different products according to the user information and the product information. For example, the similarity between different users can be calculated according to the behavior similarity between users (such as high preference for the same type of product, which can be considered as high behavior similarity), and it can be considered that the demand degree of a user for a product corresponding to another user with high similarity to itself is higher, and the demand degree for a product of the same type as the preferred product type of the user is also higher. Accordingly, the second demand degree of the target user for each target product is obtained. This can improve the possibility of pushing the product corresponding to the user with similar preferences to the target user, and improve the pushing effect.

[0069] S250, determining a third demand degree of the target product for the target user according to the user feature data and the preset at least one evaluation dimension.

[0070] S260, determining a target demand degree of the target product for the target user based on the first demand degree, the second demand degree, the third demand degree and the corresponding weight value.

[0071] The technical scheme of the embodiment, by respectively extracting features from the historical behavior data and the user feature data, obtains the historical behavior matrix and the user feature matrix which can adapt to the model input, and then inputs the user feature matrix corresponding to the user feature data into the prediction model corresponding to the integrated algorithm to obtain the first demand degree of the target product for the target user; the historical behavior matrix corresponding to the historical behavior data is input into the collaborative filtering model to obtain the second demand degree of the target product for the target user, thereby improving the data processing efficiency and the convenience of product recommendation.

[0072] Embodiment three

[0073] Figure 3 is a flowchart of a data processing method according to the embodiment three of the present application, which is further refined to S140 on the basis of the foregoing embodiment. The specific implementation can be referred to the technical scheme of the embodiment. Among them, the same or corresponding technical terms as the above embodiments are not repeated here.

[0074] As shown in Figure 3 , the method specifically comprises the following steps:

[0075] S310, obtaining historical behavior data and user feature data of a target user.

[0076] S320, processing the user feature data based on an integrated algorithm to obtain a first demand degree of a target product for the target user.

[0077] S330, performing collaborative filtering processing on the historical behavior data to determine a second demand degree of the target product for the target user.

[0078] S340, calling a preset evaluation table.

[0079] The evaluation table includes at least one evaluation dimension, and at least one evaluation item and a corresponding to-be-selected evaluation value associated with each evaluation dimension. The number of evaluation items on each evaluation dimension can be the same or different, and can be set according to actual conditions. For example, the education level information (evaluation dimension 1) can correspond to two evaluation items, such as excellent and good, and each evaluation item can correspond to a corresponding evaluation value, such as 0.6 for excellent and 0.4 for good; the income level information (evaluation dimension 2) can correspond to five evaluation items; the work type information (evaluation dimension 3) can correspond to eight evaluation items, and so on.

[0080] Specifically, an evaluation table can be established according to user behavior information and product information and other related data, and the evaluation table can be stored in a corresponding position. When a product suitable for a target user is pushed, the evaluation table can be called from a preset position, and the target user can be evaluated based on the information in the evaluation table.

[0081] S350, determining at least one target evaluation item corresponding to at least one target evaluation dimension corresponding to the user feature data.

[0082] Specifically, after obtaining the user feature data of the target user, at least one target evaluation dimension corresponding to the user feature data and at least one target evaluation item corresponding to each target evaluation dimension can be determined from the at least one evaluation item corresponding to each evaluation dimension, so that the target user can be effectively evaluated by selecting a corresponding evaluation value in each target evaluation item in combination with the user feature data.

[0083] S360, determining a target evaluation value based on a to-be-selected evaluation value corresponding to the target evaluation item.

[0084] In actual application, the to-be-selected evaluation value corresponding to each target evaluation item can be determined with reference to the user feature data. For example, assuming that the work type information in the user feature data is financial, the corresponding to-be-selected evaluation value is 0.6, assuming that the work type information in the user feature data is catering, the corresponding to-be-selected evaluation value is 0.3, and assuming that the work type information in the user feature data is mechanical, the corresponding to-be-selected evaluation value is 0.1. The determined to-be-selected evaluation value can be used as the target evaluation value. Correspondingly, the target evaluation value corresponding to each target evaluation item can be obtained.

[0085] S370, determining a third demand degree of the target product for the target user based on the target evaluation value.

[0086] In practical applications, the target user's third degree of demand for the target product can be determined by combining the target evaluation value corresponding to each target evaluation item. For example, the target evaluation values ​​can be weighted and averaged to obtain the evaluation mean. The evaluation mean can be processed to obtain a value in the range of 0-1, which can be used as the third degree of demand.

[0087] It should be noted that the third degree of demand can be determined based on human experience. For example, an expert can manually select the corresponding evaluation value from each target evaluation item in a pre-set evaluation table, referring to user characteristic data, to assess the target user's needs. Alternatively, it can be determined based on a pre-set human experience judgment model. This model is constructed based on the individual experience of experts. For instance, a user characteristic matrix corresponding to the user characteristic data can be input into the human experience judgment model, or several key indicators of interest to the expert can be pre-determined, and the columns of data corresponding to the key indicators in the user characteristic matrix can be input into the human experience judgment model to output the corresponding third degree of demand, thus improving the convenience of data processing.

[0088] S380. Based on the first demand degree, the second demand degree, the third demand degree and the corresponding weight values, determine the target user's target demand degree for the target product.

[0089] The technical solution of this embodiment retrieves a pre-set evaluation table, and then, based on user characteristic data, determines the target evaluation item corresponding to at least one target evaluation dimension in the evaluation table. Based on the candidate evaluation value corresponding to the target evaluation item, a target evaluation value is determined. Based on the target evaluation value, the third degree of demand for the target product by the target user is determined, so as to realize the evaluation of the user characteristic data of the target user by human experience method, improve the accuracy of the evaluation, and determine the target demand degree by combining the third demand degree obtained by human experience method with the first demand degree and the second demand degree. Based on the target demand degree, a target product that matches the target user is selected, thereby improving the recommendation effect.

[0090] Example 4

[0091] Figure 4 This is a flowchart of a data processing method according to Embodiment 4 of the present invention. Based on the foregoing embodiments, the method further includes updating the weight values ​​corresponding to each demand level. Specific implementation details can be found in the technical solution of this embodiment. Technical terms that are the same as or corresponding to those in the above embodiments will not be repeated here.

[0092] like Figure 4 As shown, the method specifically includes the following steps:

[0093] S410. Obtain historical behavioral data and user characteristic data of the target user.

[0094] S420, processing the user feature data based on an integration algorithm to obtain a first demand degree of the target product by the target user.

[0095] S430, performing collaborative filtering processing on the historical behavior data to determine a second demand degree of the target product by the target user.

[0096] S440, determining a third demand degree of the target product by the target user according to the user feature data and at least one evaluation dimension set in advance.

[0097] S450, determining a target demand degree of the target product by the target user based on the first demand degree, the second demand degree, the third demand degree, and corresponding weight values.

[0098] S460, determining the membership relationship between each target product and the target user within a preset time length.

[0099] The membership relationship can be used to represent the transaction behavior between the target product and the target user. For example, there is a membership relationship between product A and the target user, which means that the target user and product A have a transaction behavior. There is no membership relationship between product B and the target user, which means that the target user and product B have no transaction behavior.

[0100] In order to improve the accuracy of the recommendation, the weight values corresponding to the first demand degree, the second demand degree and the third demand degree can be updated based on the actual push feedback information. At this time, the membership relationship between each target product and the target user within a preset time length can be obtained, so as to update the weight values of each demand degree based on the products belonging to the target user and the products not belonging to the target user.

[0101] S470, determining a target value based on the membership relationship and a target function.

[0102] It should be noted that the parameter values required by the target function can be determined through the actual push information and the actual push feedback information, so that the parameter values are used as the input parameters of the target function to obtain the target value, so as to determine the weight values of each demand degree through the target value.

[0103] Optionally, before determining the target value based on the membership and the target function, further comprising: determining a first quantity of items according to the target product pushed and the product belonging to the target user in the membership; determining a second quantity of items according to the target product not pushed and the product belonging to the target user in the membership; determining a third quantity of items according to the target product not pushed and the product not belonging to the target user in the membership; determining a fourth quantity of items according to the target product not pushed and the product not belonging to the target user in the membership, so as to determine the target value based on the first quantity of items, the second quantity of items and the third quantity of items.

[0104] Wherein, the product belonging to the target user in the membership can represent that the target user possesses the target product, and the product not belonging to the target user in the membership can represent that the target user does not possess the target product.

[0105] In actual application, the quantity of the product pushed for the target user and belonging to the target user can be determined as the first quantity of items according to the target product pushed and the product belonging to the target user in the membership. The quantity of the product pushed for the target user but not belonging to the target user can be determined as the second quantity of items according to the target product pushed and the product not belonging to the target user in the membership. The quantity of the product not pushed for the target user and belonging to the target user can be determined as the third quantity of items according to the target product not pushed and the product belonging to the target user in the membership. The quantity of the product not pushed for the target user and not belonging to the target user can be determined as the fourth quantity of items according to the target product not pushed and the product not belonging to the target user in the membership, so as to determine the target value based on the first quantity of items, the second quantity of items and the third quantity of items. For example, the recall rate can be determined according to the first quantity of items and the third quantity of items, and the precision rate can be determined according to the first quantity of items and the second quantity of items. The target value can be determined based on the precision rate, the recall rate and the target function.

[0106] Wherein, the recall rate can represent the proportion of the correct prediction as positive to the overall actual positive. The precision rate can represent the proportion of the correct prediction as positive to the overall prediction as positive.

[0107] It should be noted that in specific application, the adjustment of the weight values corresponding to the first demand degree, the second demand degree and the third demand degree of the user for the product will affect the target demand degree of the user for the product, and further affect the comparison between the target demand degree and the preset demand degree threshold, affect the information of the product pushed to the user, that is, affect the first quantity of items, the second quantity of items, the third quantity of items and the fourth quantity of items, and further affect the calculated precision rate and recall rate, and affect the output result of the target function.

[0108] In the embodiment, the ratio of the first item quantity to the sum of the first item quantity and the third item quantity can be used as the recall rate. The ratio of the first item quantity to the sum of the first item quantity and the second item quantity can be used as the precision rate. For example, the first item quantity is p, the second item quantity is q, and the third item quantity is m. The recall rate r is p / (p+m), and the precision rate pr is p / (p+q). Further, the objective function can be as follows:

[0109]

[0110] wherein β is an adjustment parameter. When β is greater than 1, the adjustment effect of the change of the recall rate r in the objective function is more sensitive, and vice versa. The value of β can be determined according to the actual working condition. It should be noted that the selection of the objective function aims to balance the final result between the recall rate and the precision rate. Therefore, F is in the range of 0-1, and it can be easily seen that when β*r is close to pr, the value of F is higher, that is, the evaluation effect is better. The precision rate and the recall rate can be used as the input parameters of the objective function. The output result of the objective function will change with the change of the values of the parameters (such as the three weight values, the precision rate, or the recall rate). The weight values corresponding to the first demand degree, the second demand degree, and the third demand degree can be adjusted. When the output result of the objective function reaches the maximum, the parameter values at this time can be used as the target values, which are considered to be the optimal estimation values of the model. The model under the estimation values can achieve better prediction effect in practical application, and the model evaluation effect is the best.

[0111] S480, based on the target value, updating the weight values corresponding to the demand degrees to process the historical behavior data and the user feature data obtained next time to determine the target demand degrees of the next target user to the target products.

[0112] In practical application, the weight parameter values corresponding to the maximum output result of the objective function can be used as the new weight values to process the historical behavior data and the user feature data obtained next time to determine the target demand degrees of the next target user to the target products, so as to achieve the effect that the recommended products are the most suitable for the target user.

[0113] ​It should be noted that when the weight values corresponding to each demand degree are updated based on the target value, the weight values corresponding to each demand degree can be continuously adjusted to maximize the output result of the target function, and the weight values corresponding to each demand degree when the output result is maximized can be used as new weight values. Alternatively, updating the weight values corresponding to each demand degree based on the target value includes: determining the update percentage of the weight values corresponding to each demand degree according to the target value; and adjusting the weight values based on the corresponding update percentages.

[0114] In actual applications, when the weight values corresponding to each demand degree are adjusted, the output result of the target function can be calculated, the weight values corresponding to each demand degree when the output result is maximized can be used as the target value, and then the weight values of each demand degree can be normalized to obtain the update percentage of each demand degree, such as 40% for the first demand degree, 40% for the second demand degree, and 20% for the third demand degree, so that the sum of the three weight values is 1. The update percentages can be used as new weight values. When the first demand degree, the second demand degree, and the third demand degree of the next target user for a certain target product are obtained by processing the historical behavior data and the user feature data obtained next time, the target demand degree of the next target user for the target product can be determined based on the new weight values.

[0115] The technical scheme of the embodiment determines the membership relationship between each target product and the target user within a preset time period, determines the target value based on the membership relationship and the target function, and then updates the weight values corresponding to each demand degree based on the target value. The historical behavior data and the user feature data obtained next time are processed to determine the target demand degree of the next target user for each target product, which realizes the combination of the actually recommended product information and the actually recommended feedback information, dynamically adjusts the weight values, improves the accuracy of the target demand degree determination, further improves the adaptability of the recommended target product to the target user determined based on the target demand degree, improves the recommendation effect, and meets the user demand.

[0116] Embodiment Five

[0117] As an optional embodiment of the above embodiments, in order to make the technical scheme of the embodiments of the present application more clear to those skilled in the art, specific application scene examples are given. Specifically, refer to the following specific content.

[0118] Exemplarily, the technical solution of the embodiment combines the integrated algorithm, the collaborative filtering algorithm and the artificial experience recommendation method by weighting, and finally generates a comprehensive recommendation coefficient as the target demand degree of the user for the product. First, the user data can be collected under the premise of user permission and guarantee of user privacy, such as historical behavior data of the user and user feature data. After obtaining these data information, the data information can be processed and arranged, such as determining a historical behavior matrix corresponding to the historical behavior data, the rows of the matrix correspond to each user, and the columns correspond to the transaction records of the user for each product. If the user i and the product j exist transaction records, the element of the i-th row and the j-th column is 1, otherwise 0. Determine the user feature matrix corresponding to the user feature data, the rows of the matrix correspond to each customer, and the columns correspond to each behavior feature of the customer. It should be noted that in order to further improve the data processing efficiency, when processing and arranging the historical behavior data and the user feature data, the two data can be cleaned respectively, such as deleting or correcting abnormal values, error data, supplementing missing values, etc. The cleaning method can use the Monte Carlo algorithm to analyze the information distribution characteristics based on other normal information associated with the missing data, find the distribution function, and then simulate the new value from the distribution function as the supplementary correction result. For example, a user lacks work information, then based on the work information simulation of other users similar to the basic information of the user, it is assumed that the work information of other users is found to be approximately normally distributed through the distribution histogram and sample experience distribution function analysis, then a normal distribution random number generator can be used to simulate the work information of the user. If the data distribution of other users is too complex to be simulated by the distribution function, the Markov Chain Monte Carlo algorithm (MCMC) can be used to simulate the correction estimate value.

[0119] On the basis of the above scheme, the user feature matrix can be input into a prediction model corresponding to an integrated algorithm to obtain a first demand degree of the target user for the target product, wherein the integrated algorithm can be lightGBM, XgBoost, random forest, etc. and can be selected by the user as needed. For example, when the computing resources are insufficient, a relatively simple model such as random forest can be used, and when the computing resources are sufficient, lightGBM and XgBoost algorithms are recommended, which can improve the operation speed without reducing the accuracy, greatly saving the calculation time. Random forest, lightGBM and XgBoost algorithms can also be integrated for use. The historical behavior matrix corresponding to the historical behavior data can be input into a preset model corresponding to the integrated algorithm, and the model is fitted to obtain the model parameters and determine the model structure. The preset model can use the demand information of the user for the product as the target feature of the model, that is, the prediction result, and the target feature with demand is 1 and the target feature without demand is 0. It should be noted that the integrated algorithm can use a binary classification problem prediction mode to perform binary classification prediction on the result, and the prediction result is a value in the range of 0-1, and the larger the value, the higher the probability of the predicted customer purchasing the product. This value is used as the first demand degree, denoted as p1. If random forest, lightGBM and XgBoost algorithms are integrated for use, the prediction probabilities generated by each algorithm can be weighted and averaged to obtain the final first demand degree, still denoted as p1. The historical behavior matrix can be input into a collaborative filtering model to record the behavior characteristics of each user for each product, calculate the similarity, and complete the fitting work of the collaborative filtering recommendation model. After the collaborative filtering model is fitted, it can be used for recommendation calculation, and the final output of the model can reflect the data of the demand possibility of each customer for each product. This data is processed by scaling, shifting and other data processing methods, and finally converted into a value between 0 and 1 as the second demand degree, denoted as p2. The user feature data can be processed by using the artificial experience recommendation method to obtain the third demand degree. For example, the user feature data can be used to select the corresponding evaluation value in each target evaluation item in the pre-set evaluation table to effectively evaluate the target user. The third demand degree can also be determined based on a pre-set artificial experience judgment model, wherein the artificial experience judgment model is constructed based on the personal experience of experts. For example, the user feature matrix corresponding to the user feature data can be input into the artificial experience judgment model, or the user feature matrix corresponding to the key indicators of the user feature data can be input into the artificial experience judgment model, and the corresponding third demand degree can be output, denoted as p3.

[0120] On the basis of the above scheme, the first demand degree, the second demand degree and the third demand degree of a certain target user to the product can be weighted and combined to obtain the final comprehensive preference coefficient of the target user to the product as the target demand degree. For example, the first demand degree is p1, the corresponding weight value is w1, the second demand degree is p2, the corresponding weight value is w2, and the third demand degree is p3, the corresponding weight value is w3. At this time, the target demand degree can be s=w1*p1+w2*p2+w3*p3. A demand degree threshold T can be set. When s≥T, it is determined that the demand degree of the user to the target product is high, that is, the recommendation can be made. When s<T, it is determined that the demand degree of the user to the target product is low, and no recommendation is made.

[0121] On the basis of the above scheme, the membership between each target product and the target user in the preset time period can also be determined. Then, according to the target products pushed and the products belonging to the target user in the membership, the number of products belonging to the target user and pushed to the target user is determined as the first item quantity, denoted as p. According to the target products pushed and the products not belonging to the target user in the membership, the number of products not belonging to the target user and pushed to the target user is determined as the second item quantity, denoted as q. According to the target products not pushed and the products belonging to the target user in the membership, the number of products belonging to the target user and not pushed to the target user is determined as the third item quantity, denoted as m. Further, the recall rate r=p / (p+m) and the precision rate pr=p / (p+q) are obtained. The precision rate and the recall rate can be used as the input parameters of the target function, and the target function can be:

[0122]

[0123] wherein β is an adjustment parameter. When β is greater than 1, the adjustment effect of the change of the recall rate r in the target function is more sensitive, and vice versa. It should be noted that the selection of the target function aims to balance the final result between the recall rate and the precision rate, so the value of F is in the range of 0-1, and it can be easily seen that when β*r is greater than pr, F is greater than 0.5, and vice versa. Therefore, the target function can be used to determine the target product to be pushed to the target user. The closer the values are, the larger the F value is when the values of pr and r are larger, that is, the evaluation effect is considered to be better. The precision and recall can be used as the input parameters of the objective function, and the output result of the objective function will change with the change of the values of the parameters (such as the three weight values, the precision or the recall). The weight values corresponding to the first demand degree, the second demand degree and the third demand degree can be adjusted, when the output result of the objective function reaches the maximum, the parameter values at this time can be used as the optimal estimation value of the model, the model under the estimation value can achieve better prediction effect in practical application, the model evaluation effect is best, and the parameters, the weight coefficients and the threshold values of the model algorithm are determined, so that a more objective judgment system of the given model structure is determined, and finally the recommended model is determined and applied to the real scene.

[0124] The technical scheme of the embodiment realizes the demand degrees determined by the three evaluation methods of integrated algorithm, collaborative filtering and evaluation dimension, and determines the target demand degree which can effectively and accurately represent the demand degree of the target user for the target product based on the demand degrees and the corresponding weight values, so as to determine the target product to be pushed to the target user based on the target demand degree, so that the target product determined to be pushed to the target user is matched with the target user, the recommendation effect is improved, and the technical effect of meeting the user demand is achieved.

[0125] Embodiment six

[0126] Figure 5 is a structural schematic diagram of a data processing device provided by the embodiment six of the application. As shown in Figure 5 the device comprises a data acquisition module 510, a first demand degree determination module 520, a second demand degree determination module 530, a third demand degree determination module 540 and a target demand degree determination module 550.

[0127] The data acquisition module 510 is configured to acquire historical behavior data and user feature data of a target user. The first demand degree determination module 520 is configured to process the user feature data based on an integrated algorithm to obtain a first demand degree of the target user for a target product. The second demand degree determination module 530 is configured to perform collaborative filtering processing on the historical behavior data to determine a second demand degree of the target user for the target product. The third demand degree determination module 540 is configured to determine a third demand degree of the target user for the target product according to the user feature data and at least one pre-set evaluation dimension. The target demand degree determination module 550 is configured to determine a target demand degree of the target user for the target product based on the first demand degree, the second demand degree, the third demand degree and corresponding weight values.

[0128] The technical scheme of the embodiment is characterized in that: the historical behavior data and the user feature data of a target user are acquired; the user feature data is processed based on an integrated algorithm to obtain a first demand degree of the target user for a target product; the historical behavior data is processed by collaborative filtering to determine a second demand degree of the target user for the target product; the third demand degree of the target user for the target product is determined according to the user feature data and at least one pre-set evaluation dimension; and the target demand degree of the target user for the target product is determined based on the first demand degree, the second demand degree, the third demand degree and corresponding weight values, thereby solving the problem that in the prior art, a product with high transaction frequency in user historical transaction information is recommended to a user, resulting in poor recommendation effect and causing the recommended product to be incompatible with the user, and achieving the demand degree determined by three evaluation methods, namely, a comprehensive integrated algorithm, collaborative filtering and an evaluation dimension, and based on the demand degrees and corresponding weight values, determining the target demand degree that can effectively and accurately represent the demand degree of the target user for the target product, so as to determine the target product to be pushed to the target user based on the target demand degree, so that the target product determined to be pushed to the target user is matched with the target user, the recommendation effect is improved, and the technical effect of meeting the user demand is achieved.

[0129] On the basis of the above device, optionally, the target short message slice type determination module 520 includes a target short message template determination unit and a target short message slice type determination unit.

[0130] On the basis of the above device, optionally, the historical behavior data includes historical acquisition record information of at least one to-be-selected product, wherein the at least one to-be-selected product includes the target product, and the user feature data includes user basic data and asset information.

[0131] On the basis of the above device, optionally, the device further includes a feature extraction module, which is configured to perform feature extraction on the historical behavior data and the user feature data respectively to obtain a historical behavior matrix corresponding to the historical behavior data and a user feature matrix corresponding to the user feature data.

[0132] On the basis of the above device, optionally, the first demand degree determination module 520 includes a first demand degree determination unit.

[0133] The first demand degree determination unit is configured to input the user feature matrix corresponding to the user feature data into a prediction model corresponding to the integrated algorithm to obtain the first demand degree of the target user for the target product.

[0134] On the basis of the above device, optionally, the second demand degree determination module 530 includes a second demand degree determination unit.

[0135] The second demand degree determination unit is configured to input a historical behavior matrix corresponding to the historical behavior data into a collaborative filtering model to obtain a second demand degree of the target user for the target product.

[0136] The collaborative filtering model is determined based on a predetermined training set data matrix corresponding to a historical behavior matrix of each user to be trained.

[0137] On the basis of the above device, the third demand degree determination module 540 can include an evaluation table determination unit, a target evaluation item determination unit, a target evaluation value determination unit, and a third demand degree determination unit.

[0138] The evaluation table determination unit is configured to call a pre-set evaluation table, wherein the evaluation table includes at least one evaluation dimension, at least one evaluation item associated with each evaluation dimension, and a corresponding to-be-selected evaluation value.

[0139] The target evaluation item determination unit is configured to determine, according to the user feature data, a target evaluation item corresponding to at least one target evaluation dimension corresponding thereto.

[0140] The target evaluation value determination unit is configured to determine, based on a to-be-selected evaluation value corresponding to the target evaluation item, a target evaluation value.

[0141] The third demand degree determination unit is configured to determine, based on the target evaluation value, a third demand degree of the target user for the target product.

[0142] On the basis of the above device, the target demand degree determination module 550 can include a first intermediate value determination unit, a second intermediate value determination unit, a third intermediate value determination unit, and a target demand degree determination unit.

[0143] The first intermediate value determination unit is configured to determine a first intermediate value according to a product of the first demand degree and a corresponding weight value.

[0144] The second intermediate value determination unit is configured to determine a second intermediate value based on a product of the second demand degree and a corresponding weight value.

[0145] The third intermediate value determination unit is configured to determine a third intermediate value according to a product of the third demand degree and a corresponding weight value.

[0146] The target demand degree determination unit is configured to determine the target demand degree based on the first intermediate value, the second intermediate value, and the third intermediate value.

[0147] On the basis of the above device, optionally, the device further comprises a product pushing module, the product pushing module is used for if the target demand degree is greater than a preset demand degree threshold, pushing the target product to the target user;

[0148] The product pushing module is further used for if the target demand degree is less than a preset demand degree threshold, not pushing the target product to the target user.

[0149] On the basis of the above device, optionally, the device further comprises a weight value updating module, the weight value updating module comprises a membership relationship determining unit, a target value determining unit and a weight value updating determining unit.

[0150] The membership relationship determining unit is used for determining the membership relationship between each target product and the target user within a preset time length;

[0151] The target value determining unit is used for determining a target value based on the membership relationship and a target function;

[0152] The weight value updating determining unit is used for updating the weight value corresponding to each demand degree based on the target value, so as to process the historical behavior data and the user feature data obtained next time, and determine the target demand degree of the next target user to each target product.

[0153] On the basis of the above device, optionally, the weight value updating module further comprises a first article quantity determining unit, a second article quantity determining unit, a third article quantity determining unit and a fourth article quantity determining unit.

[0154] The first article quantity determining unit is used for determining a first article quantity according to the target product pushed and the product belonging to the target user in the membership relationship;

[0155] The second article quantity determining unit is used for determining a second article quantity according to the target product pushed and the product not belonging to the target user in the membership relationship;

[0156] The third article quantity determining unit is used for determining a third article quantity according to the target product not pushed and the product belonging to the target user in the membership relationship;

[0157] The fourth article quantity determining unit is used for determining a fourth article quantity according to the target product not pushed and the product not belonging to the target user in the membership relationship, so as to determine the target value based on the first article quantity, the second article quantity and the third article quantity.

[0158] On the basis of the above device, optionally, the target value determination unit comprises a recall rate determination subunit, a precision rate determination subunit and a target value determination subunit.

[0159] The recall rate determination subunit is configured to determine a recall rate according to the first item quantity and the third item quantity.

[0160] The precision rate determination subunit is configured to determine a precision rate according to the first item quantity and the second item quantity.

[0161] The target value determination subunit is configured to determine the target value based on the precision rate, the recall rate and a target function.

[0162] On the basis of the above device, optionally, the weight value update determination unit comprises an update percentage determination subunit and a weight value update subunit.

[0163] The update percentage determination subunit is configured to determine an update percentage of each weight value corresponding to the demand degree according to the target value.

[0164] The weight value update subunit is configured to adjust each weight value based on the corresponding update percentage.

[0165] The data processing device provided by the embodiment of the present application can execute the data processing method provided by any embodiment of the present application, and has the corresponding function modules and beneficial effects of the execution method.

[0166] Embodiment seven

[0167] Figure 6 is a structural schematic diagram of an electronic device for implementing the data processing method of the embodiment of the present application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections, and relationships, and their functions, are merely examples and are not intended to limit the implementations described and / or claimed in this document.

[0168] As Figure 6As shown, the electronic device 10 includes at least one processor 11, and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., communicatively connected to the at least one processor 11, where the memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0169] Various components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc., an output unit 17, such as various types of displays, a speaker, etc., a storage unit 18, such as a magnetic disk, an optical disk, etc., and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0170] The processor 11 can be various general and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 performs various methods and processes described above, such as a data processing method.

[0171] In some embodiments, the data processing method can be implemented as a computer program tangibly embodied in a computer readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded onto the RAM 13 and executed by the processor 11, one or more steps of the data processing method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the data processing method by any other appropriate means, such as by means of firmware.

[0172] The various embodiments of the systems and techniques described above can be implemented in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a load programmable logic device (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0173] Computer programs used to implement the processes of the application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the computer program

[0174] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store computer programs for use by or in connection with an instruction execution system, apparatus, or device. Computer-readable storage media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium will include one or more lines of electrical connections, portable computer disks, hard disk drives, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0175] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0176] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0177] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. A server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0178] It should be understood that the various forms of flow shown above can be re-ordered, added to, or deleted from without departing from the scope of the present application. For example, the steps recited in the present application can be executed in parallel, in series, or in a different order, without departing from the desired results of the technical solutions of the present application, and the present application is not limited herein.

[0179] Embodiment Eight

[0180] The embodiment of the present application further provides a computer program product, comprising a computer program which, when executed by a processor, implements the data processing method provided in any embodiment of the present application.

[0181] Computer program products can be written in any one of a number of programming languages, including object oriented languages such as Java, Smalltalk, C++, or conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0182] The specific embodiments have been shown and described for purposes of illustrating the present application. It will be readily apparent to those skilled in the art that various modifications, combinations, sub-combinations, and alternatives can be made to the specific embodiments without departing from the spirit and principles of the present application. Accordingly, the present application is not intended to be limited to the specific embodiments described herein.

Claims

1. A data processing method, characterized by, The method comprises: obtaining historical behavior data and user feature data of a target user; processing the user feature data based on an integrated algorithm to obtain a first demand degree of the target user for a target product; performing collaborative filtering processing on the historical behavior data to determine a second demand degree of the target user for the target product; determining a third demand degree of the target user for the target product according to the user feature data and at least one pre-set evaluation dimension; determining a target demand degree of the target user for the target product based on the first demand degree, the second demand degree, the third demand degree, and corresponding weight values. The method further comprises: determining a membership relationship between each target product and the target user within a preset time period; determining a first number of items according to the target products pushed and the products belonging to the target user in the membership relationship; determining a second number of items according to the target products pushed and the products not belonging to the target user in the membership relationship; determining a third number of items according to the target products not pushed and the products belonging to the target user in the membership relationship; determining a fourth number of items according to the target products not pushed and the products not belonging to the target user in the membership relationship, to determine a target value based on the first number of items, the second number of items, and the third number of items; determining the target value based on the membership relationship and a target function, comprising: determining a recall rate according to the first number of items and the third number of items; determining a precision rate according to the first number of items and the second number of items; determining the target value based on the precision rate, the recall rate, and the target function; updating the weight values corresponding to each demand degree based on the target value, comprising: determining an update percentage of the weight values corresponding to each demand degree based on the target value, and adjusting the weight values based on the corresponding update percentages; processing the historical behavior data and the user feature data obtained next time to determine target demand degrees of a next target user for each target product.

2. The method of claim 1, wherein, The historical behavior data includes historical acquisition record information of at least one product to be selected, wherein the at least one product to be selected includes the target product, and the user feature data includes user basic data and asset information.

3. The method of claim 1, wherein, After obtaining the historical behavior data and the user feature data of the target user, the method further comprises: performing feature extraction on the historical behavior data and the user feature data respectively to obtain a historical behavior matrix corresponding to the historical behavior data and a user feature matrix corresponding to the user feature data.

4. The method of claim 1, wherein, The processing of the user feature data based on the integrated algorithm to obtain the first demand degree of the target user for the target product comprises: inputting the user feature matrix corresponding to the user feature data into a prediction model corresponding to the integrated algorithm to obtain the first demand degree of the target user for the target product.

5. The method of claim 1, wherein, The collaborative filtering processing of the historical behavior data to determine the second demand degree of the target user for the target product comprises: inputting a historical behavior matrix corresponding to the historical behavior data into a collaborative filtering model to obtain a second demand degree of the target user for the target product; The collaborative filtering model is determined based on a predetermined training set data matrix corresponding to a historical behavior matrix of each user to be trained.

6. The method of claim 1, wherein, The third demand degree of the target user for the target product is determined according to the user feature data and at least one evaluation dimension set in advance, including: The evaluation table includes at least one evaluation dimension, at least one evaluation item associated with each evaluation dimension, and a corresponding to-be-selected evaluation value; According to the user feature data, at least one target evaluation dimension corresponding to the target evaluation item is determined; Based on the to-be-selected evaluation value corresponding to the target evaluation item, a target evaluation value is determined; Based on the target evaluation value, the third demand degree of the target user for the target product is determined.

7. The method of claim 1, wherein, The target demand degree of the target user for the target product is determined based on the first demand degree, the second demand degree, the third demand degree, and the corresponding weight value, including: A first intermediate value is determined according to the product of the first demand degree and the corresponding weight value; A second intermediate value is determined based on the product of the second demand degree and the corresponding weight value; A third intermediate value is determined according to the product of the third demand degree and the corresponding weight value; The target demand degree is determined based on the first intermediate value, the second intermediate value, and the third intermediate value.

8. The method of claim 1, wherein, Further comprising: If the target demand degree is greater than a preset demand degree threshold, the target product is pushed to the target user; If the target demand degree is less than a preset demand degree threshold, the target product is not pushed to the target user.

9. A data processing apparatus, characterized by, Comprising: The data acquisition module is used for acquiring historical behavior data and user feature data of a target user; The first demand degree determination module is used for processing the user feature data based on an integrated algorithm to obtain a first demand degree of the target user for a target product; The second demand degree determination module is used for collaborative filtering processing of the historical behavior data to determine a second demand degree of the target user for the target product; The third demand degree determination module is used for determining a third demand degree of the target user for the target product according to the user feature data and at least one evaluation dimension set in advance; The target demand degree determination module is used for determining a target demand degree of the target user for the target product based on the first demand degree, the second demand degree, the third demand degree, and the corresponding weight value. The device further comprises a weight value updating module, which comprises: The membership relationship determination unit is used for determining the membership relationship between each target product and the target user within a preset time length; The first item quantity determination unit is used for determining a first item quantity according to the target product pushed and the product belonging to the target user in the membership relationship; The second item quantity determination unit is used for determining a second item quantity according to the target product pushed and the product not belonging to the target user in the membership relationship; A third item quantity determining unit is configured to determine a third item quantity according to the target product not pushed and the product belonging to the target user in the affiliation relationship; A fourth item quantity determining unit is configured to determine a fourth item quantity according to the target product not pushed and the product not belonging to the target user in the affiliation relationship, so as to determine a target value based on the first item quantity, the second item quantity and the third item quantity; A target value determining unit is configured to determine the target value based on the affiliation relationship and a target function; The target value determining unit comprises: A recall rate determining sub-unit is configured to determine a recall rate according to the first item quantity and the third item quantity; A precision rate determining sub-unit is configured to determine a precision rate according to the first item quantity and the second item quantity; A target value determining sub-unit is configured to determine the target value based on the precision rate, the recall rate and the target function; A weight value updating determining unit is configured to update the weight value corresponding to each demand degree based on the target value, so as to process the historical behavior data and the user feature data obtained next time and determine the target demand degree of each target product for the next target user; The weight value updating determining unit comprises An updating percentage determining sub-unit is configured to determine an updating percentage of the weight value corresponding to each demand degree according to the target value; A weight value updating sub-unit is configured to adjust each weight value based on the corresponding updating percentage.

10. An electronic device, comprising: The computer program is executed by the processor to implement the data processing method of any one of claims 1-8.

11. A computer readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the data processing method of any one of claims 1-8.

12. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the data processing method of any one of claims 1-8.

Images