A data processing method, device and storage medium

Abstract

The embodiment of the present application provides a data processing method and device, and a storage medium, the method comprising: extracting object attribute information, historical behavior data corresponding to the object attribute information, and historical display data from historical log data; and obtaining historical recommendation data corresponding to the object attribute information from a historical recommendation information library; searching for first historical recommendation data that is identical to the historical display data from the historical recommendation data; and obtaining second historical recommendation data based on the historical display data, the historical behavior data, and the first historical recommendation data; training a preset recommendation model using the second historical recommendation data and the third historical recommendation data to obtain a trained preset recommendation model; and upon receiving first identity attribute information, determining recommendation data corresponding to the first identity attribute information based on the trained preset recommendation model.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a data processing method and device, and a storage medium. Background Art

[0002] Due to the large size of the product library, in order to meet the demand for real-time product recommendations to customers, currently, a recommendation system is usually used to filter user-related products from the product library based on user attribute information, and then recommend them to the user. However, during the model training process, the existing recommendation system has a low accuracy due to the singleness of the samples. Summary of the invention

[0003] The embodiments of the present application provide a data processing method and device, and a storage medium, which can improve the accuracy of recommendations.

[0004] The technical solution of this application is implemented as follows:

[0005] In a first aspect, an embodiment of the present application provides a data processing method, the method comprising:

[0006] Extracting object attribute information, historical behavior data and historical display data corresponding to the object attribute information from historical log data; and acquiring historical recommendation data corresponding to the object attribute information from a historical recommendation information library; the historical recommendation data includes the historical display data;

[0007] Searching the historical recommendation data for first historical recommendation data that is the same as the historical display data; and obtaining second historical recommendation data according to the historical display data, the historical behavior data and the first historical recommendation data;

[0008] Using the second historical recommendation data and the third historical recommendation data to train a preset recommendation model to obtain a trained preset recommendation model; the third historical recommendation data is the historical recommendation data other than the first historical recommendation data in the historical recommendation data;

[0009] When the first identity attribute information is received, the recommended data corresponding to the first identity attribute information is determined based on the trained preset recommendation model.

[0010] In the above data processing method, obtaining the second historical recommendation data according to the historical display data, the historical behavior data and the first historical recommendation data includes:

[0011] Performing data splicing on the historical display data and the first historical recommendation data to obtain recommendation display data;

[0012] dividing the recommendation display data into recommendation display click data and recommendation display non-click data according to the historical behavior data; and determining the recommendation display click data and the recommendation display non-click data as the second historical recommendation data;

[0013] Correspondingly, the third historical recommendation data is recommended undisplayed data, and the method of training a preset recommendation model using the second historical recommendation data and the third historical recommendation data to obtain a trained preset recommendation model includes:

[0014] The preset recommendation model is trained using the recommendation display click data, the recommendation display non-click data and the recommendation non-display data to obtain a trained preset recommendation model.

[0015] In the above data processing method, the using of the recommendation display click data, the recommendation display non-click data and the recommendation non-display data to train a preset recommendation model includes:

[0016] Inputting each of the recommendation display click data, the recommendation display non-click data and the recommendation non-display data into a preset recommendation model in sequence to obtain a predicted recommendation display click rate, a predicted recommendation display non-click rate and a predicted recommendation non-display rate corresponding to each of the data;

[0017] A preset recommendation model is trained based on the predicted recommendation display click rate, the predicted recommendation display non-click rate and the predicted recommendation non-display rate.

[0018] In the above data processing method, determining the recommended data corresponding to the first identity attribute information based on the trained preset recommendation model includes:

[0019] Searching for the to-be-recommended data corresponding to the first identity attribute information from a target database;

[0020] Inputting the data to be recommended into the trained preset recommendation model to obtain the recommendation display click rate, recommendation display non-click rate and recommendation non-display rate corresponding to the data to be recommended;

[0021] Determine the recommended data corresponding to the first identity attribute information from the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate, and the recommendation non-display rate.

[0022] In the above data processing method, determining the recommended data corresponding to the first identity attribute information from the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate includes:

[0023] Determine the recommendation index of the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate;

[0024] Sorting the data to be recommended according to the order of the to-be-recommended index from high to low to obtain sorted data to be recommended;

[0025] A preset number of data to be recommended is selected from the sorted data to be recommended, and the preset number of data to be recommended is determined as the recommended data corresponding to the first identity attribute information.

[0026] In a second aspect, an embodiment of the present application provides a data processing device, the device comprising:

[0027] an acquisition unit, configured to extract object attribute information, historical behavior data and historical display data corresponding to the object attribute information from historical log data; and to acquire historical recommendation data corresponding to the object attribute information from a historical recommendation information library; the historical recommendation data includes the historical display data;

[0028] a search unit, configured to search the historical recommendation data for first historical recommendation data identical to the historical display data; and obtain second historical recommendation data according to the historical display data, the historical behavior data and the first historical recommendation data;

[0029] a model training unit, configured to train a preset recommendation model using the second historical recommendation data and the third historical recommendation data to obtain a trained preset recommendation model; the third historical recommendation data is the historical recommendation data other than the first historical recommendation data in the historical recommendation data;

[0030] A determination unit is used to determine the recommendation data corresponding to the first identity attribute information based on the trained preset recommendation model when receiving the first identity attribute information.

[0031] In the above data processing device, the device further comprises: an input unit;

[0032] The searching unit is further used to search the target database for the to-be-recommended data corresponding to the first identity attribute information;

[0033] The input unit is used to input the data to be recommended into the trained preset recommendation model to obtain the recommendation display click rate, recommendation display non-click rate and recommendation non-display rate corresponding to the data to be recommended;

[0034] The determining unit is further configured to determine the recommended data corresponding to the first identity attribute information from the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate, and the recommendation non-display rate.

[0035] In the above data processing device, the device further comprises: a sorting unit;

[0036] The determining unit is further used to determine the recommendation index of the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate;

[0037] The sorting unit is used to sort the data to be recommended in descending order of the to-be-recommended indexes to obtain sorted data to be recommended;

[0038] The determining unit is further configured to select a preset number of data to be recommended from the sorted data to be recommended, and determine the preset number of data to be recommended as the recommended data corresponding to the first identity attribute information.

[0039] In a third aspect, an embodiment of the present application provides a data processing device, characterized in that the device comprises: a processor, a memory and a communication bus; when the processor executes an operating program stored in the memory, the data processing method as described in any one of the above items is implemented.

[0040] In a fourth aspect, an embodiment of the present application provides a storage medium on which a computer program is stored, characterized in that when the computer program is executed by a processor, it implements the data processing method as described in any one of the above items.

[0041] The embodiment of the present application provides a data processing method and device, and a storage medium, the method comprising: extracting object attribute information, historical behavior data corresponding to the object attribute information, and historical display data from historical log data; and obtaining historical recommendation data corresponding to the object attribute information from a historical recommendation information library; the historical recommendation data comprises historical display data; searching for first historical recommendation data identical to the historical display data from the historical recommendation data; and obtaining second historical recommendation data based on the historical display data, the historical behavior data, and the first historical recommendation data; training a preset recommendation model using the second historical recommendation data and the third historical recommendation data to obtain a trained preset recommendation model; the third historical recommendation data is the historical recommendation data except the first historical recommendation data. historical recommendation data other than the first identity attribute information; when the first identity attribute information is received, the recommendation data corresponding to the first identity attribute information is determined based on the trained preset recommendation model; the above-mentioned implementation scheme is adopted to obtain the second historical recommendation data and the third historical recommendation data by extracting and concatenating the historical log data and the historical recommendation data, and then use the second historical recommendation data and the third historical recommendation data to train the preset recommendation model. Through the technical scheme of the present application, since sample data of multiple dimensions are used when training the preset recommendation model, not only a single type of sample data is considered, the accuracy of training the preset recommendation model can be improved, thereby improving the accuracy of prediction using the trained preset recommendation model. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 A schematic diagram of an exemplary recommended device provided for an embodiment of the present application;

[0043] Figure 2 A flow chart of a data processing method provided in an embodiment of the present application;

[0044] Figure 3 A schematic diagram of an exemplary data processing device provided in an embodiment of the present application;

[0045] Figure 4 A schematic diagram of the structure of a data processing device provided in an embodiment of the present application;

[0046] Figure 5 A schematic diagram of the composition structure of a data processing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0047] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0048] This application is based on an improvement of a recommended device. Figure 1 An exemplary schematic diagram of a recommended device is given in the embodiment of the present application, such as Figure 1 As shown, the recommendation device includes three modules: a recall module, a rough sorting module and a fine sorting module.

[0049] The recall module is used to find the products related to the user from the product library using different recall methods, and return the recall results to the rough sorting.

[0050] The coarse sorting module is used to use the trained coarse sorting click-through rate prediction model to predict the click-through rate of the obtained recall results, and after sorting the recall results according to the prediction results, return the products ranked in the top preset number to the fine sorting module, for example, return the products ranked in the top 100 to the fine sorting module.

[0051] The fine sorting module, in the same way as the rough sorting module, is used to select the final recommended products from a preset number of products and display them on the display interface for user operation.

[0052] However, in the existing recommendation device, since only the historical display data displayed on the user display interface is used for training the rough ranking click rate prediction model, the problem of sample selection bias is caused, which leads to low prediction accuracy of the rough ranking click rate prediction model.

[0053] Based on the above technical problems, the present application proposes a data processing method, which can solve the sample bias problem in the existing recommendation device and thus improve the accuracy of the recommendation.

[0054] The present application provides a data processing method, which is applied to a data processing device. Figure 2 A flow chart of a data processing method provided in an embodiment of the present application is as follows: Figure 2 As shown, the data processing method may include:

[0055] S101, extracting object attribute information, historical behavior data corresponding to the object attribute information, and historical display data from historical log data; and acquiring historical recommendation data corresponding to the object attribute information from a historical recommendation information database.

[0056] The data processing method proposed in the embodiment of the present application can be applied in the scenario of recommending products to users.

[0057] In an embodiment of the present application, the data processing device extracts object attribute information, historical behavior data corresponding to the object attribute information, and historical display data from historical log data; and obtains historical recommendation data corresponding to the object attribute information from a historical recommendation information library.

[0058] refer to Figure 3 , Figure 3A schematic diagram of an exemplary data processing device proposed in an embodiment of the present application, wherein the data processing device is compared with Figure 1 In terms of the system, a recall log module is added between the recall module and the coarse sorting module, and a sorting strategy is added to the coarse sorting module. It should be noted that the recall log module is used to obtain historical recommendation data from the recall module, and generate historical display data and historical behavior data, and then process the historical recommendation data, historical display data and historical behavior data to generate training samples for training the preset recommendation model. The sorting strategy is used to sort the recommended data to filter out the recommended data during the data recommendation process using the trained preset recommendation model.

[0059] Specifically, the recall log module is composed of a log generation module, a log collection module, a log parsing module and a training data generation module, wherein the log generation module is used to obtain object attribute information, historical behavior data and historical display data from historical log data; the log collection module is used to obtain historical recommendation data corresponding to the object attribute information from the historical recommendation information library.

[0060] It should be noted that the object attribute information may be the user's attribute information, such as the user's identity document (ID), user age, and user location. The specific object attribute information is determined based on actual conditions and is not limited in the embodiments of the present application.

[0061] It should be noted that the historical behavior data may be the historical behavior data of a user (object), such as the user's historical browsing data, historical query data, historical purchase data, historical purchase data, and historical comments, etc. The specific historical behavior data is determined based on actual conditions and is not limited in the embodiments of the present application.

[0062] It should be noted that the historical display data may be data that is historically displayed on the user's display interface. For example, after user A opens a shopping app, the product data displayed to user A on the display interface of the shopping app is the historical display data. The specific historical display data is determined based on actual conditions and is not limited in this embodiment of the present application.

[0063] It should be noted that the historical display data includes product data, such as the product number corresponding to the product, the product category and the product price. The specific product data is determined according to the actual situation and is not limited in the embodiment of the present application.

[0064] In the embodiment of the present application, the historical recommendation data in the historical recommendation information library is obtained by Figure 2It is generated by the recall module in. Recall can be understood as retrieving user-related information from the material library based on user behavior, portrait and other information. There can be multiple recall methods, such as "region-based recall" and "age-based recall". The products in the target database are scored respectively by the two recall methods, and finally the scores obtained by the two methods are merged to obtain the final score. At this time, the products with scores greater than the preset values ​​are historical recommendation data; the specific preset values ​​and recall methods are determined according to actual conditions, and the embodiments of the present application are not limited here.

[0065] It should be noted that the historical recommendation data includes the historical display data. Specifically, the recommendation data historically displayed on the display interface in the historical recommendation data is the historical display data.

[0066] S102, searching the historical recommendation data for first historical recommendation data that is the same as the historical display data; and obtaining second historical recommendation data according to the historical display data, the historical behavior data and the first historical recommendation data.

[0067] In an embodiment of the present application, after obtaining the historical recommendation data, the data processing device searches the historical recommendation data for the first historical recommendation data that is the same as the historical display data; and obtains the second historical recommendation data based on the historical display data, the historical behavior data and the first historical recommendation data.

[0068] It should be noted that Figure 3 The log parsing module in the recall log module is used to obtain the second historical recommendation data according to the historical display data, the historical behavior data and the first historical recommendation data.

[0069] It should be noted that the second historical recommendation data is composed of recommendation display click data and recommendation display non-click data.

[0070] Specifically, the historical display data and the first historical recommendation data are concatenated to obtain the recommendation display data; the recommendation display data is divided into recommendation display click data and recommendation display non-click data according to the historical behavior data; and the recommendation display click data and the recommendation display non-click data are determined as the second historical recommendation data.

[0071] It should be noted that the recommended display data can be obtained by concatenating the historical display data and the first historical recommendation data, and then the historical behavior data is used to obtain which data in the recommended display data has been clicked, thereby dividing the recommended display data into recommended display click data and recommended display non-click data.

[0072] It can be understood that by dividing the recommendation display data into recommendation display click data and recommendation display non-click data according to historical behavior data, and then using the recommendation display click data and recommendation display non-click data to train the preset recommendation model, the accuracy of model training is improved due to the increase in the data type during training.

[0073] S103: Use the second historical recommendation data and the third historical recommendation data to train a preset recommendation model to obtain a trained preset recommendation model.

[0074] In an embodiment of the present application, after obtaining the second historical recommendation data, the data processing device uses the second historical recommendation data and the third historical recommendation data to train a preset recommendation model to obtain a trained preset recommendation model.

[0075] In an embodiment of the present application, after obtaining the second historical recommendation data (recommendation display click data and recommendation display non-click data), the data processing device needs to first obtain the third historical recommendation data (recommendation non-display data) from the historical recommendation data.

[0076] In the embodiment of the present application, the third historical recommendation data is the historical recommendation data other than the first historical recommendation data;

[0077] It should be noted that, since a part of the historical recommendation data is displayed as historical display data, the remaining part of the historical recommendation data is historical recommendation data that is not displayed, that is, recommended non-displayed data.

[0078] In an embodiment of the present application, after obtaining the recommendation display click data, the recommendation display non-click data and the recommendation non-display data, the data processing device uses the recommendation display click data, the recommendation display non-click data and the recommendation non-display data to train a preset recommendation model to obtain the trained preset recommendation model.

[0079] It should be noted that the preset recommendation model may be a click-through rate prediction (CTR) model. The specific preset recommendation model is determined according to actual conditions and is not limited in the embodiments of the present application.

[0080] Specifically, each piece of recommended display click data, recommended display non-click data and recommended non-display data is input into the preset recommendation model in turn to obtain the predicted recommendation display click rate, predicted recommendation display non-click rate and predicted recommendation non-display rate corresponding to each piece of data; the preset recommendation model is trained based on the predicted recommendation display click rate, predicted recommendation display non-click rate and predicted recommendation non-display rate.

[0081] It should be noted that for these three types of training data, the embodiment of the present application chooses to use the cross entropy loss function for training in the process of training the preset recommendation model. In practical applications, other loss functions can also be used for model training, which is not limited in the embodiment of the present application.

[0082] It should be noted that when the data is input into the preset recommendation model, the output result is three values. For example, the output result can be (predicted recommendation display click rate 50%, predicted recommendation display non-click rate 30% and predicted recommendation non-display rate 20%), and the sum of the three values ​​is always 1.

[0083] In an embodiment of the present application, before using the recommendation display click data, the recommendation display not click data and the recommendation not displayed data to train the preset recommendation model, a preset proportion of the first recommendation display click data, the first recommendation display not click data and the first recommendation not displayed data can also be selected from the recommendation display click data, the recommendation display not click data and the recommendation not displayed data to train the preset recommendation model.

[0084] In the embodiment of the present application, by Figure 3 The log parsing module of the recall log module obtains the recommended display click data, recommended display non-click data, and recommended non-displayed data. At this time, the number of data may be uneven, which will also affect the final model training results. Therefore, it is necessary to Figure 3 The training data generation module of the recall log module extracts a preset proportion of data from the recommendation display click data, the recommendation display non-click data and the recommendation non-display data to train the preset recommendation model. For example, assuming that 100 recommendation display click data, 200 recommendation display non-click data and 300 recommendation non-display data are obtained at this time, and the preset ratio is set to 1:1:1, at this time, it is necessary to select 100 data from the 200 recommendation display non-click data and the 300 recommendation non-display data respectively, and use them together with the 100 recommendation display click data as samples to train the preset recommendation model; the specific preset ratio is determined according to the actual situation, and the embodiment of the present application is not limited here.

[0085] It should be noted that Figure 3 The training data generation module in the recall log module is used to extract features from the object attribute information obtained by the log generation module, and use the extracted attribute features to train the preset recommendation model, thereby improving the accuracy of the prediction of the preset recommendation model after training.

[0086] It can be understood that when training the preset recommendation model, not only the second historical recommendation data (recommendation display click data and recommendation display non-click data) is obtained, but also the third historical recommendation data (recommendation non-display data) is obtained from the historical recommendation data, and then the three types of data obtained are input into the preset recommendation model for training. The preset recommendation model can be comprehensively trained using data from multiple dimensions, thereby improving the prediction rate of the preset recommendation model.

[0087] S104: When the first identity attribute information is received, determine the recommended data corresponding to the first identity attribute information based on the trained preset recommendation model.

[0088] In an embodiment of the present application, after completing the training of the preset recommendation model, upon receiving the first identity attribute information, the data processing device determines the recommended data corresponding to the first identity attribute information based on the trained preset recommendation model.

[0089] It should be noted that the first identity attribute information is the information required for a user to log in to a data processing device, such as account information or a mobile phone number, etc. The data processing device may be a device for deploying application software.

[0090] It should be noted that Figure 3 The rough sorting module is used to use the trained preset recommendation model to determine the recommended data corresponding to the first identity attribute information.

[0091] Specifically, when the first identity attribute information is received, the data to be recommended corresponding to the first identity attribute information is searched from the target database; the data to be recommended is input into the trained preset recommendation model to obtain the recommendation display click rate, recommendation display non-click rate and recommendation non-display rate corresponding to the data to be recommended; the recommendation data corresponding to the first identity attribute information is determined from the data to be recommended according to the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate.

[0092] It should be noted that the target database is Figure 3 In the product library in the target database, products related to the first identity attribute information are searched according to the first identity attribute information as the data to be recommended.

[0093] It should be noted that after obtaining the data to be recommended, each piece of data in the data to be recommended is input into the trained preset recommendation model in turn to obtain the output result corresponding to each piece of data, namely, the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate. Finally, the recommended data is determined from the data to be recommended according to the output results and then input into the recommended model. Figure 3 In the fine sorting module shown, the results are finally displayed on the display interface.

[0094] It should be noted that Figure 3 The sorting strategy in the coarse sorting module is used to sort the data to be recommended, and then determine the recommended data from the sorted data to be recommended.

[0095] Specifically, the recommendation index of the data to be recommended is determined according to the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate; the data to be recommended is sorted in descending order of the recommendation index to obtain the sorted data to be recommended; a preset number of data to be recommended is selected from the sorted data to be recommended, and the preset number of data to be recommended is determined as the recommendation data corresponding to the first identity attribute information.

[0096] In the embodiment of the present application, the recommendation display click rate is recorded as P1, the recommendation display non-click rate is recorded as P2, and the recommendation non-display rate is recorded as P3. Before calculating the recommendation index, it is necessary to first calculate the recommendation display rate P4 as P1+P2, and then obtain the recommendation index through the following formula:

[0097] S=P1 t1 ×P4 t2 (1)

[0098] In the above formula (1), t1 is the first parameter configured for P1, and t2 is the second parameter configured for P4.

[0099] It should be noted that, since the recommendation non-display rate P3 is considered to have no possibility of being clicked in actual situations, the to-be-recommended data with a P3 value higher than the preset threshold can be placed at the back. Afterwards, for the to-be-recommended data with a P3 value lower than the preset threshold, the recommendation index is calculated according to the method of formula (1) and the to-be-recommended data with a P3 value lower than the preset threshold is sorted in front according to the recommendation index.

[0100] It should be noted that P1 and P2 can also be used directly to calculate the recommendation index of the recommended data. Whether to use P3 can be selected according to the actual situation, and the embodiment of the present application does not limit it here. For example, suppose there are five recommended products A, B, C, D and E, and their recommendation indexes are 5, 4, 7, 8 and 3 respectively. After sorting A, B, C, D and E according to the recommendation index, the sorting results are D, C, A, B and E. It is necessary to select the products with the top 4 scores as recommended products. At this time, the recommended products are D, C, A and B; the specific preset number is determined according to the actual situation, and the embodiment of the present application does not limit it here.

[0101] It can be understood that since the preset recommendation model is trained with three different types of data: recommendation display click data, recommendation display non-click data, and recommendation non-display data, it is able to improve the accuracy of recommending the first identity attribute information using the preset recommendation model, thereby improving the user's click-through rate.

[0102] An embodiment of the present application provides a data processing method, which includes: extracting object attribute information, historical behavior data and historical display data corresponding to the object attribute information from historical log data; and obtaining historical recommendation data corresponding to the object attribute information from a historical recommendation information library; the historical recommendation data includes historical display data; searching for first historical recommendation data that is the same as the historical display data from the historical recommendation data; and obtaining second historical recommendation data based on the historical display data, the historical behavior data and the first historical recommendation data; training a preset recommendation model using the second historical recommendation data and the third historical recommendation data to obtain a trained preset recommendation model; the third historical recommendation data is the historical recommendation data in the historical recommendation data other than the first historical recommendation data; when receiving the first identity attribute information, determining the recommendation data corresponding to the first identity attribute information based on the trained preset recommendation model; using the above implementation scheme, by extracting and splicing the historical log data and the historical recommendation data, obtaining multiple types of sample data, and then using multiple types of sample data to train the preset recommendation model can improve the accuracy of prediction using the trained model, thereby improving the accuracy of recommendation.

[0103] Based on the above embodiment, in another embodiment of the present application, a data processing device 1 is provided. Figure 4 A schematic diagram of the structure of a data processing device provided in this application is shown in FIG. Figure 4 As shown, the data processing device 1 includes:

[0104] The acquisition unit 10 is used to extract object attribute information, historical behavior data and historical display data corresponding to the object attribute information from the historical log data; and to acquire the historical recommendation data corresponding to the object attribute information from the historical recommendation information library; the historical recommendation data includes the historical display data;

[0105] A search unit 11 is used to search the historical recommendation data for first historical recommendation data that is the same as the historical display data; and obtain second historical recommendation data according to the historical display data, the historical behavior data and the first historical recommendation data;

[0106] A model training unit 12 is used to train a preset recommendation model using the second historical recommendation data and the third historical recommendation data to obtain a trained preset recommendation model; the third historical recommendation data is the historical recommendation data other than the first historical recommendation data in the historical recommendation data;

[0107] The determination unit 13 is used to determine the recommended data corresponding to the first identity attribute information based on the trained preset recommendation model when receiving the first identity attribute information.

[0108] Optionally, the data processing device 1 further includes: a data processing unit;

[0109] The data processing unit is used to perform data splicing on the historical display data and the first historical recommendation data to obtain recommendation display data;

[0110] The data processing unit is further configured to divide the recommendation display data into recommendation display click data and recommendation display non-click data according to the historical behavior data; and determine the recommendation display click data and the recommendation display non-click data as the second historical recommendation data;

[0111] The model training unit 12 is further used to train a preset recommendation model using the recommendation display click data, the recommendation display non-click data and the recommendation non-display data to obtain a trained preset recommendation model.

[0112] Optionally, the data processing device 1 further includes: an input unit;

[0113] The input unit is used to input each of the recommendation display click data, the recommendation display non-click data and the recommendation non-display data into a preset recommendation model in sequence, and obtain the predicted recommendation display click rate, the predicted recommendation display non-click rate and the predicted recommendation non-display rate corresponding to each of the data;

[0114] The model training unit 12 is further used to train a preset recommendation model based on the predicted recommendation display click rate, the predicted recommendation display non-click rate and the predicted recommendation non-display rate.

[0115] Optionally, the searching unit 11 is further configured to search for the to-be-recommended data corresponding to the first identity attribute information from a target database;

[0116] The input unit is further used to input the data to be recommended into the trained preset recommendation model to obtain the recommendation display click rate, recommendation display non-click rate and recommendation non-display rate corresponding to the data to be recommended;

[0117] The determining unit 13 is further configured to determine the recommended data corresponding to the first identity attribute information from the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate.

[0118] Optionally, the data processing device 1 further includes: a sorting unit;

[0119] The determining unit 13 is further configured to determine a recommendation index of the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate;

[0120] The sorting unit is used to sort the data to be recommended in descending order of the to-be-recommended indexes to obtain sorted data to be recommended;

[0121] The determining unit 13 is further configured to select a preset number of data to be recommended from the sorted data to be recommended, and determine the preset number of data to be recommended as the recommended data corresponding to the first identity attribute information.

[0122] An embodiment of the present application provides a data processing device, which includes: extracting object attribute information, historical behavior data and historical display data corresponding to the object attribute information from historical log data; and obtaining historical recommendation data corresponding to the object attribute information from a historical recommendation information library; the historical recommendation data includes historical display data; searching for first historical recommendation data that is the same as the historical display data from the historical recommendation data; and obtaining second historical recommendation data based on the historical display data, the historical behavior data and the first historical recommendation data; training a preset recommendation model using the second historical recommendation data and the third historical recommendation data to obtain a trained preset recommendation model; the third historical recommendation data is the historical recommendation data in the historical recommendation data other than the first historical recommendation data; when receiving the first identity attribute information, determining the recommendation data corresponding to the first identity attribute information based on the trained preset recommendation model; using the above implementation scheme, by extracting and splicing the historical log data and the historical recommendation data, obtaining multiple types of sample data, and then using multiple types of sample data to train the preset recommendation model can improve the accuracy of prediction using the trained model, thereby improving the accuracy of recommendation.

[0123] Figure 5 A schematic diagram of the structure of a data processing device provided in an embodiment of the present application. In practical applications, based on the same disclosure concept of the above embodiments, Figure 5 As shown, the data processing device 2 of this embodiment includes: a processor 20 , a memory 21 and a communication bus 22 .

[0124] In the process of a specific embodiment, the acquisition unit 10, the search unit 11, the model training unit 12, the determination unit 13, the data processing unit, the input unit and the sorting unit can be implemented by a processor 20 located on the data processing device 2, and the processor 20 can be at least one of an application-specific integrated circuit (ASIC, Application Specific Integrated Circuit), a digital signal processor (DSP, Digital Signal Processor), a digital signal processing image processing device (DSPD, Digital Signal Processing Device), a programmable logic image processing device (PLD, Programmable Logic Device), a field programmable gate array (FPGA, Field Programmable Gate Array), a CPU, a controller, a microcontroller, and a microprocessor. It can be understood that for different data processing devices, the electronic device used to implement the above-mentioned processor function can also be other, and this embodiment does not specifically limit it.

[0125] In the embodiment of the present application, the communication bus 22 is used to realize the connection and communication between the processor 20 and the memory 21; when the processor 20 executes the running program stored in the memory 21, the following data processing method is implemented:

[0126] Extracting object attribute information, historical behavior data and historical display data corresponding to the object attribute information from historical log data; and acquiring historical recommendation data corresponding to the object attribute information from a historical recommendation information library; the historical recommendation data includes the historical display data;

[0127] Searching the historical recommendation data for first historical recommendation data that is the same as the historical display data; and obtaining second historical recommendation data according to the historical display data, the historical behavior data and the first historical recommendation data;

[0128] Using the second historical recommendation data and the third historical recommendation data to train a preset recommendation model to obtain a trained preset recommendation model; the third historical recommendation data is the historical recommendation data other than the first historical recommendation data in the historical recommendation data;

[0129] When the first identity attribute information is received, the recommended data corresponding to the first identity attribute information is determined based on the trained preset recommendation model.

[0130] Optionally, the processor 20 is also used to concatenate the historical display data and the first historical recommendation data to obtain recommended display data; divide the recommended display data into recommended display click data and recommended display non-click data according to the historical behavior data; and determine the recommended display click data and the recommended display non-click data as the second historical recommendation data; accordingly, the third historical recommendation data is recommended non-displayed data, and using the second historical recommendation data and the third historical recommendation data to train a preset recommendation model to obtain a trained preset recommendation model, includes: using the recommended display click data, the recommended display non-click data and the recommended non-displayed data to train a preset recommendation model to obtain a trained preset recommendation model.

[0131] Optionally, the processor 20 is also used to input each of the recommendation display click data, the recommendation display non-click data and the recommendation non-display data into a preset recommendation model in turn, to obtain the predicted recommendation display click rate, the predicted recommendation display non-click rate and the predicted recommendation non-display rate corresponding to each of the data; and to train the preset recommendation model based on the predicted recommendation display click rate, the predicted recommendation display non-click rate and the predicted recommendation non-display rate.

[0132] Optionally, the processor 20 is also used to search for the data to be recommended corresponding to the first identity attribute information from the target database; input the data to be recommended into the trained preset recommendation model to obtain the recommendation display click rate, recommendation display non-click rate and recommendation non-display rate corresponding to the data to be recommended; determine the recommendation data corresponding to the first identity attribute information from the data to be recommended according to the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate.

[0133] Optionally, the processor 20 is also used to determine the recommendation index of the data to be recommended based on the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate; sort the data to be recommended in order from high to low according to the recommendation index to obtain sorted data to be recommended; select a preset number of data to be recommended from the sorted data to be recommended, and determine the preset number of data to be recommended as the recommended data corresponding to the first identity attribute information.

[0134] An embodiment of the present application provides a storage medium having a computer program stored thereon. The computer-readable storage medium stores one or more programs. The one or more programs can be executed by one or more processors and applied to a data processing device. The computer program implements the data processing method as described above.

[0135] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0136] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present disclosure, or the part that contributes to the relevant technology, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling an image display device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the data processing method described in each embodiment of the present disclosure.

[0137] The above description is only a preferred embodiment of the present application and is not intended to limit the protection scope of the present application.

Claims

1. A data processing method, characterized in that: The method comprises: Extracting user attribute information, historical behavior data and historical display data corresponding to the user attribute information from historical log data; and acquiring historical recommendation data corresponding to the user attribute information from a historical recommendation information library; the historical recommendation data includes the historical display data; wherein the historical display data represents data historically displayed on the display interface of the user; Searching the historical recommendation data for first historical recommendation data that is the same as the historical display data; Performing data splicing on the historical display data and the first historical recommendation data to obtain recommendation display data; dividing the recommendation display data into recommendation display click data and recommendation display non-click data according to the historical behavior data; and determining the recommendation display click data and the recommendation display non-click data as second historical recommendation data; Using the second historical recommendation data and the third historical recommendation data to train a preset recommendation model to obtain a trained preset recommendation model; the third historical recommendation data is the historical recommendation data other than the first historical recommendation data in the historical recommendation data; When the first identity attribute information is received, the recommended data corresponding to the first identity attribute information is determined based on the trained preset recommendation model.

2. The method according to claim 1, characterized in that The third historical recommendation data is recommended undisplayed data, and the using the second historical recommendation data and the third historical recommendation data to train a preset recommendation model to obtain a trained preset recommendation model includes: The preset recommendation model is trained using the recommendation display click data, the recommendation display non-click data and the recommendation non-display data to obtain a trained preset recommendation model.

3. The method according to claim 2, characterized in that The using the recommendation display click data, the recommendation display non-click data and the recommendation non-display data to train a preset recommendation model includes: Inputting each of the recommendation display click data, the recommendation display non-click data and the recommendation non-display data into a preset recommendation model in sequence to obtain a predicted recommendation display click rate, a predicted recommendation display non-click rate and a predicted recommendation non-display rate corresponding to each of the data; A preset recommendation model is trained based on the predicted recommendation display click rate, the predicted recommendation display non-click rate and the predicted recommendation non-display rate.

4. The method according to claim 1, characterized in that: The determining the recommended data corresponding to the first identity attribute information based on the trained preset recommendation model includes: Searching for the to-be-recommended data corresponding to the first identity attribute information from a target database; Inputting the data to be recommended into the trained preset recommendation model to obtain the recommendation display click rate, recommendation display non-click rate and recommendation non-display rate corresponding to the data to be recommended; Determine the recommended data corresponding to the first identity attribute information from the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate, and the recommendation non-display rate.

5. The method according to claim 4, characterized in that The determining, from the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate, and the recommendation non-display rate, the recommendation data corresponding to the first identity attribute information includes: Determine the recommendation index of the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate; Sorting the data to be recommended according to the order of the recommendation indexes of the data to be recommended from high to low to obtain sorted data to be recommended; A preset number of data to be recommended is selected from the sorted data to be recommended, and the preset number of data to be recommended is determined as the recommended data corresponding to the first identity attribute information.

6. A data processing device, characterized in that: The device comprises: an acquisition unit, configured to extract user attribute information, historical behavior data corresponding to the user attribute information, and historical display data from historical log data; and to acquire historical recommendation data corresponding to the user attribute information from a historical recommendation information library; the historical recommendation data includes the historical display data; wherein the historical display data represents data historically displayed on a display interface of the user; a search unit, configured to search the historical recommendation data for first historical recommendation data identical to the historical display data; perform data splicing on the historical display data and the first historical recommendation data to obtain recommendation display data; divide the recommendation display data into recommendation display click data and recommendation display non-click data according to the historical behavior data; and determine the recommendation display click data and the recommendation display non-click data as second historical recommendation data; a model training unit, configured to train a preset recommendation model using the second historical recommendation data and the third historical recommendation data to obtain a trained preset recommendation model; the third historical recommendation data is the historical recommendation data other than the first historical recommendation data in the historical recommendation data; A determination unit is used to determine the recommendation data corresponding to the first identity attribute information based on the trained preset recommendation model when receiving the first identity attribute information.

7. The device according to claim 6, characterized in that The device further comprises: an input unit; The searching unit is further used to search the target database for the to-be-recommended data corresponding to the first identity attribute information; The input unit is used to input the data to be recommended into the trained preset recommendation model to obtain the recommendation display click rate, recommendation display non-click rate and recommendation non-display rate corresponding to the data to be recommended; The determining unit is further configured to determine the recommended data corresponding to the first identity attribute information from the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate, and the recommendation non-display rate.

8. The device according to claim 7, characterized in that The device further comprises: a sorting unit; The determining unit is further used to determine the recommendation index of the to-be-recommended data according to the recommendation display click rate, the recommendation display non-click rate and the recommendation non-display rate; The sorting unit is used to sort the data to be recommended in descending order of the recommendation index of the data to be recommended to obtain sorted data to be recommended; The determining unit is further configured to select a preset number of data to be recommended from the sorted data to be recommended, and determine the preset number of data to be recommended as the recommended data corresponding to the first identity attribute information.

9. A data processing device, characterized in that: The device comprises: a processor, a memory and a communication bus; when the processor executes the running program stored in the memory, the method according to any one of claims 1 to 5 is implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images