Content recommendation method and device, equipment, storage medium and product

Abstract

The embodiment of the invention relates to the field of financial science and technology, and particularly provides a content recommendation method and device, equipment, a storage medium and a product, and the method comprises the steps: obtaining target content corresponding to a to-be-recommended user, and a target content type corresponding to the to-be-recommended user; performing module division on the target content according to a plurality of modules corresponding to the target content type to obtain a plurality of modules corresponding to the target content, the plurality of modules reflecting information in the target content from a plurality of different angles; performing module division on the target content to obtain a plurality of modules of the target content; according to the browsing behavior data and the click behavior data of the to-be-recommended user, determining at least one target module in the target content corresponding to the user and the priority of the at least one target module; and configuring the at least one target module based on the priority, and recommending the configured target module to the to-be-recommended user. According to the embodiment of the invention, the content recommendation accuracy can be improved.

Description

Technical Field

[0001] The embodiments of this specification relate to the field of financial technology, and in particular, to a content recommendation method, device, equipment, storage medium and product. Background Art

[0002] With the continuous development of science and technology and information technology, in the digital age, more and more content, such as advertisements, products, services, activities, etc., need to be recommended and displayed to users. The existing content recommendation method may directly provide a large amount of content to all users without distinction. In this process, some users may be interested in the content products provided, while others may not be interested.

[0003] The existing content recommendation methods are unable to make targeted content recommendations based on the needs and value perceptions of different user groups towards products.

[0004] Therefore, there is an urgent need for a content recommendation method that can perform targeted content recommendations and improve the accuracy of content recommendations. Summary of the invention

[0005] The purpose of the embodiments of this specification is to provide a content recommendation method, apparatus, device, storage medium and product to improve the accuracy of content recommendation.

[0006] To achieve the above objectives, on the one hand, an embodiment of this specification provides a content recommendation method, including:

[0007] Obtain the target content corresponding to the user to be recommended and the target content type corresponding to the user to be recommended;

[0008] Divide the target content into modules according to the modules corresponding to the target content type to obtain the modules corresponding to the target content, wherein the modules reflect the information in the target content from different perspectives;

[0009] According to the historical browsing behavior data and historical clicking behavior data of the user to be recommended, the historical preference score of each module in the target content type corresponding to the user to be recommended and the historical behavior frequency of all modules in the target content type corresponding to the user to be recommended are calculated;

[0010] According to the current browsing behavior data and current click behavior data of the user to be recommended, the current preference score of each module in the target content type corresponding to the user to be recommended and the current behavior frequency of all modules in the target content type corresponding to the user to be recommended are calculated;

[0011] Compare the historical behavior frequency with the current behavior frequency, and based on the comparison result, directly or indirectly sort the modules according to the historical preference scores and the current preference scores;

[0012] Obtaining at least one target module and its priority corresponding to the user to be recommended according to the direct sorting result or the indirect sorting result;

[0013] At least one target module in the target content is configured based on the priority, and the configured target module is recommended to a user to be recommended.

[0014] Preferably, the step of obtaining target content corresponding to the user to be recommended further includes:

[0015] Determining whether the user to be recommended is a specific user according to the user information of the user to be recommended;

[0016] If yes, determining the target content corresponding to the user to be recommended;

[0017] If not, extracting the user features of the user to be recommended from the associated information of the user to be recommended;

[0018] The user characteristics are input into the trained neural network model to obtain the target content corresponding to the user to be recommended.

[0019] Preferably, the specific users include whitelist users and tagged users.

[0020] Preferably, extracting the user characteristics of the user to be recommended from the associated information of the user to be recommended further comprises:

[0021] Extracting age characteristics, gender characteristics, location characteristics, education level characteristics and occupation characteristics from the user information of the user to be recommended;

[0022] Extracting device type features and operating system type features from the device information of the user to be recommended;

[0023] Extracting account login time features, account level features, and account payment preference features from the account information of the user to be recommended;

[0024] The consumption type characteristics, consumption frequency characteristics and consumption amount characteristics are extracted from the consumption information of the user to be recommended.

[0025] Preferably, the target content type is at least one of an activity recommendation type, a product recommendation type, an advertisement recommendation type, and a service recommendation type.

[0026] Preferably, dividing the target content into modules according to the modules corresponding to the target content type to obtain the modules corresponding to the target content further comprises:

[0027] If the target content type is an activity recommendation type, then according to several modules corresponding to the activity recommendation type, the target content is divided into an activity introduction module, a discount content module and a participation process module;

[0028] If the target content type is a product recommendation type, then according to several modules corresponding to the product recommendation type, the target content is divided into a product introduction module, a discount content module, a user evaluation module and a purchase process module;

[0029] If the target content type is an advertisement recommendation type, the target content is divided into an advertisement introduction module, a discount content module, a user evaluation module, and an action guidance module according to several modules corresponding to the advertisement recommendation type;

[0030] If the target content type is a service recommendation type, the target content is divided into a service introduction module, a preferential content module, a user evaluation module and a participation process module according to several modules corresponding to the loan product type.

[0031] Preferably, the calculating of the historical preference scores of each module in the target content type corresponding to the user to be recommended and the historical behavior frequencies of all modules in the target content type corresponding to the user to be recommended based on the historical browsing behavior data and the historical clicking behavior data of the user to be recommended further comprises:

[0032] Extract the historical browsing times, historical average browsing time and historical post-browsing operations of each module in the target content type according to the historical browsing behavior data of the user to be recommended;

[0033] Counting the number of first operations of clicks after the historical browsing in each module, and calculating a first ratio of the first operation number to the historical browsing number in each module;

[0034] Extracting the historical click counts and historical post-click operations of each module in the target content type according to the historical click behavior data of the user to be recommended;

[0035] Counting the number of second operations in each module that are effective conversions after the historical click operation, and calculating a second ratio of the second operation number to the historical click number in each module;

[0036] Standardizing the historical browsing times, historical average browsing duration, and first ratio of each module to obtain the processed historical browsing times, processed historical browsing duration, and processed first ratio of each module;

[0037] Performing weighted calculation on the processed historical browsing times, processed historical browsing durations and processed first ratios of each module to obtain a historical browsing preference score for each module;

[0038] Standardize the historical number of clicks and the second ratio of each module to obtain the processed historical number of clicks and the processed second ratio of each module;

[0039] Performing weighted calculation on the processed historical click counts and the processed second ratio of each module to obtain the historical click preference score of each module;

[0040] The historical browsing preference scores and historical click preference scores of each module are combined to obtain the historical preference scores of each module;

[0041] According to the total number of historical views of all modules, the historical average viewing frequency is calculated;

[0042] According to the total number of historical clicks of all modules, the historical average click frequency is calculated;

[0043] Calculate the average of the historical average browsing frequency and the historical average click frequency as the historical behavior frequency of all modules in the target content type corresponding to the recommended user.

[0044] Preferably, the standardizing the historical browsing times, historical average browsing duration and first ratio of each module to obtain the processed historical browsing times, processed historical browsing duration and processed first ratio of each module further comprises:

[0045] Based on the historical browsing times, historical average browsing durations, and first ratios of each module in each module in multiple different historical time periods, multiple historical browsing times, multiple historical average browsing durations, and multiple first ratios of each module are obtained;

[0046] According to the maximum and minimum values ​​of the multiple historical browsing times of each module, the multiple historical browsing times of each module are standardized;

[0047] Calculate the average of multiple standardized historical browsing times as the historical browsing times after each module is processed;

[0048] According to the maximum and minimum values ​​of the multiple historical average browsing durations of each module, the multiple historical average browsing durations of each module are standardized;

[0049] Calculate the average of the historical average browsing time after multiple standardization processes as the historical browsing time after each module is processed;

[0050] According to the maximum value and the minimum value of the multiple first ratios of each module, the multiple first ratios of each module are standardized;

[0051] Calculate the average of the first proportions after multiple standardization processes as the first proportion after each module in each module;

[0052] The standardizing of the historical click counts and the second ratio of each module to obtain the processed historical click counts and the processed second ratio of each module further includes:

[0053] Based on the historical number of clicks and the second ratio of each module in each module in multiple different historical time periods, multiple historical number of clicks and multiple second ratios of each module are obtained;

[0054] According to the maximum and minimum values ​​of the multiple historical click counts of each module, the multiple historical click counts of each module are standardized;

[0055] Calculate the average of multiple standardized historical click counts as the historical click count after each module is processed;

[0056] According to the maximum value and the minimum value of the multiple second ratios of each module, the multiple second ratios of each module are standardized;

[0057] An average value of the second ratios after the multiple standardization processes is calculated as the second ratio after the processing of each module in each module.

[0058] Preferably, the step of calculating the current preference score of each module in the target content type corresponding to the user to be recommended and the current behavior frequency of all modules in the target content type corresponding to the user to be recommended based on the current browsing behavior data and the current click behavior data of the user to be recommended further comprises:

[0059] Extracting the current browsing times, current average browsing time and current post-browsing operations of each module in the target content type according to the current browsing behavior data of the user to be recommended;

[0060] Counting the number of third operations in each module in which the operation after the current browsing is click, and calculating the third ratio of the number of third operations in each module to the current browsing number;

[0061] Extracting the current click count and current post-click operations of each module in the target content type according to the current click behavior data of the user to be recommended;

[0062] Counting the number of fourth operations in each module in which the current post-click operation is a valid conversion, and calculating the fourth ratio of the fourth operation number in each module to the current click number;

[0063] Standardizing the current browsing times, current average browsing time and third ratio of each module to obtain the processed current browsing times, processed current browsing time and processed third ratio of each module;

[0064] Performing weighted calculation on the processed current browsing times, processed current browsing duration and processed third ratio of each module to obtain the current browsing preference score of each module;

[0065] Performing standardization processing on the current number of clicks and the fourth ratio of each module to obtain the processed current number of clicks and the processed fourth ratio of each module;

[0066] Performing weighted calculation on the processed current click counts and the processed fourth ratio of each module to obtain the current click preference score of each module;

[0067] The current browsing preference score and the current clicking preference score of each module are integrated to obtain the current preference score of each module;

[0068] According to the sum of the current browsing times of all modules, the current average browsing frequency is calculated;

[0069] According to the sum of the current click times of all modules, the current average click frequency is calculated;

[0070] Calculate the average of the current average browsing frequency and the current average clicking frequency as the current behavior frequency of all modules in the target content type corresponding to the recommended user.

[0071] Preferably, the standardization of the current browsing times, the current average browsing time and the third ratio of each module to obtain the processed current browsing times, the processed current browsing time and the processed third ratio of each module further includes:

[0072] Based on the current browsing times, current average browsing durations and third ratios of each module in each module in multiple different current time periods, multiple current browsing times, multiple current average browsing durations and multiple third ratios of each module are obtained;

[0073] According to the maximum and minimum values ​​of the multiple current browsing times of each module, the multiple current browsing times of each module are standardized;

[0074] Calculate the average of the current page views after multiple standardization processes as the current page views after each module in each module;

[0075] According to the maximum and minimum values ​​of the multiple current average browsing durations of each module, the multiple current average browsing durations of each module are standardized;

[0076] Calculate the average of the current average browsing time after multiple standardization processes as the current browsing time after each module in each module;

[0077] According to the maximum value and the minimum value of the multiple third ratios of each module, the multiple third ratios of each module are normalized;

[0078] Calculate the average of the third proportions after multiple standardization processes as the third proportion after each module in each module;

[0079] The standardization of the current number of clicks and the fourth ratio of each module to obtain the processed current number of clicks and the processed fourth ratio of each module further includes:

[0080] Based on the current number of clicks and the fourth ratio of each module in each module in multiple different current time periods, multiple current number of clicks and multiple fourth ratios of each module are obtained;

[0081] According to the maximum and minimum values ​​of the multiple current click times of each module, the multiple current click times of each module are standardized;

[0082] Calculate the average of the current click counts after multiple standardization processes as the current click count after each module in each module;

[0083] According to the maximum value and the minimum value of the plurality of fourth ratios of each module, the plurality of fourth ratios of each module are normalized;

[0084] An average value of the fourth ratios after the multiple standardization processes is calculated as the fourth ratio after the processing of each module in each module.

[0085] Preferably, the comparing the historical behavior frequency and the current behavior frequency, and based on the comparison result, directly or indirectly sorting according to the historical preference score and the current preference score of each module further comprises:

[0086] Comparing the historical behavior frequency with the current behavior frequency;

[0087] If the historical behavior frequency and the current behavior frequency are equal, then the modules are directly sorted according to their historical preference scores and current preference scores;

[0088] If the historical behavior frequency and the current behavior frequency are not equal, an indirect sorting is performed based on the historical preference scores and the current preference scores of each module.

[0089] Preferably, the direct sorting according to the historical preference scores and current preference scores of each module further comprises:

[0090] Sort the historical preference scores in descending order, and use the modules corresponding to the top N historical preference scores as the historical preference modules of the user to be recommended;

[0091] The sorting order corresponding to the historical preference module is used as the priority of the historical preference module;

[0092] Sort the current preference scores in descending order, and use the modules corresponding to the top N current preference scores as the current preference modules of the user to be recommended;

[0093] The sorting order corresponding to the current preference module is used as the priority of the current preference module.

[0094] Preferably, obtaining at least one target module corresponding to the to-be-recommended user and its priority according to the direct sorting result further comprises:

[0095] If the historical preference module and the current preference module of the user to be recommended are exactly the same, the exactly the same module is used as the target module, and the priority of the target module is obtained according to the average priority of the exactly the same historical preference module and the current preference module;

[0096] If the historical preference module and the current preference module of the user to be recommended are partially the same, the modules that are partially the same are used as the target module, and the priority of the target module is obtained according to the average priority of the historical preference modules and the current preference modules that are partially the same;

[0097] If the historical preference module and the current preference module of the user to be recommended are completely different, the current preference module is used as the target module, and the priority of the current preference module is used as the priority of the target module.

[0098] Preferably, the indirect sorting according to the historical preference scores and current preference scores of each module further includes:

[0099] Determine the historical weight and current weight based on the ratio of historical behavior frequency to current behavior frequency;

[0100] The historical score of each module is obtained by multiplying the historical preference score of each module by the corresponding historical weight;

[0101] The current score of each module is obtained by multiplying the current preference score of each module by the corresponding current weight;

[0102] Sum the historical scores and current scores of each module to get the final score of each module;

[0103] Sort the final scores in descending order.

[0104] Preferably, obtaining at least one target module corresponding to the to-be-recommended user and its priority according to the indirect sorting result further comprises:

[0105] The module corresponding to the final score of the top M ranked modules is taken as the target module;

[0106] The sorting order corresponding to the target modules is used as the priority of the target modules.

[0107] Preferably, configuring at least one target module in the target content based on the priority and recommending the configured target module to the user to be recommended further comprises:

[0108] Based on the priorities, target modules with higher priorities in the target content are set in the focus area of ​​the browsing field of the user to be recommended, and target modules with lower priorities are set in the non-focus area of ​​the browsing field of the user to be recommended, to obtain configured target modules;

[0109] Recommend the configured target module to the recommended user.

[0110] On the other hand, an embodiment of the present specification provides a content recommendation device, the device comprising:

[0111] An acquisition module is used to acquire the target content corresponding to the user to be recommended and the target content type corresponding to the user to be recommended;

[0112] A division module, used to divide the target content into modules according to the modules corresponding to the target content type, so as to obtain the modules corresponding to the target content, wherein the modules reflect the information in the target content from different perspectives;

[0113] A history calculation module, used to calculate the historical preference score of each module in the target content type corresponding to the user to be recommended, and the historical behavior frequency of all modules in the target content type corresponding to the user to be recommended, based on the historical browsing behavior data and historical click behavior data of the user to be recommended;

[0114] A current calculation module, used to calculate the current preference score of each module in the target content type corresponding to the user to be recommended, and the current behavior frequency of all modules in the target content type corresponding to the user to be recommended, based on the current browsing behavior data and current click behavior data of the user to be recommended;

[0115] A comparison module, used to compare the historical behavior frequency with the current behavior frequency, and based on the comparison result, directly or indirectly sort according to the historical preference score and the current preference score of each module;

[0116] A target module and priority determination module, used to obtain at least one target module and its priority corresponding to the user to be recommended according to the direct sorting result or the indirect sorting result;

[0117] A configuration module is used to configure at least one target module in the target content based on the priority, and recommend the configured target module to a user to be recommended.

[0118] On the other hand, an embodiment of the present specification further provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein when the computer program is executed by the processor, the instructions of any one of the above methods are executed.

[0119] On the other hand, an embodiment of the present specification further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor of a computer device, the computer program executes instructions of any one of the above-described methods.

[0120] On the other hand, the embodiments of this specification further provide a computer program product, which, when executed by a processor of a computer device, executes instructions of any one of the above methods.

[0121] It can be seen from the technical solutions provided by the above embodiments of this specification that the embodiments of this specification divide the target content into multiple modules through the module division method, and these modules reflect the core information of the content from different angles. Based on the user's browsing and clicking behavior data, the modules that the user is interested in are determined and prioritized. The determination of priority depends on the specific performance of the user in the historical behavior data, and the relevant information can be displayed to the user first, further improving the accuracy of the recommendation.

[0122] Configure the recommended content module according to the priority to ensure that the recommended content is in line with the user's interests and can guide the user to generate effective conversions. Flexible adjustment of content through modular configuration to match it with user needs, and maximize the accuracy of content recommendations by combining user information, behavioral data and model predictions.

[0123] In order to make the above and other purposes, features and advantages of the present specification more obvious and easy to understand, the following specifically cites preferred embodiments and describes them in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0124] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0125] Figure 1 A schematic diagram showing a flow chart of a content recommendation method provided in an embodiment of this specification;

[0126] Figure 2 A schematic diagram of a process of extracting user features of a user to be recommended from associated information of the user to be recommended provided in an embodiment of this specification is shown;

[0127] Figure 3 A schematic diagram of a process for determining a historical preference module of a user to be recommended and a priority of the historical preference module according to the historical browsing behavior data and historical click behavior data of the user to be recommended provided in an embodiment of this specification is shown;

[0128] Figure 4 A schematic diagram of a process for performing standardization processing on the historical browsing times, the historical average browsing time and the first ratio of each module provided in an embodiment of the present specification to obtain the processed historical browsing times, the processed historical browsing time and the processed first ratio of each module is shown;

[0129] Figure 5 A schematic diagram of a process for performing normalization processing on the historical click counts and the second ratio of each module to obtain the processed historical click counts and the processed second ratio of each module provided in an embodiment of this specification is shown;

[0130] Figure 6 A schematic diagram of a process for determining the current preference module of a user to be recommended and the priority of the current preference module according to the current browsing behavior data and current click behavior data of the user to be recommended provided by an embodiment of this specification is shown;

[0131] Figure 7 A schematic diagram of a process for performing standardization processing on the current browsing times, the current average browsing time and the third ratio of each module provided in an embodiment of the present specification to obtain the processed current browsing times, the processed current browsing time and the processed third ratio of each module is shown;

[0132] Figure 8 A schematic diagram of a process for performing standardization processing on the current number of clicks and the fourth ratio of each module to obtain the processed current number of clicks and the processed fourth ratio of each module provided in an embodiment of this specification is shown;

[0133] Fig. 9 A schematic diagram of a process of configuring at least one target module based on a priority and recommending the configured target module to a user to be recommended is shown in an embodiment of this specification;

[0134] Fig.10 A schematic diagram of the module structure of a content recommendation device provided in an embodiment of this specification is shown;

[0135] Fig.11 A schematic diagram of the structure of a computer device provided in an embodiment of this specification is shown.

[0136] Description of the accompanying symbols:

[0137] 100. Get module;

[0138] 200, divide the module;

[0139] 300. History calculation module;

[0140] 400, Current Computing Module;

[0141] 500, comparison module;

[0142] 600, target module and priority determination module;

[0143] 700, configuration module;

[0144] 1302. Computer equipment;

[0145] 1304, processor;

[0146] 1306. Memory;

[0147] 1308, driving mechanism;

[0148] 1310, input / output module;

[0149] 1312. Input devices;

[0150] 1314. Output device;

[0151] 1316. Presentation equipment;

[0152] 1318. Graphical user interface;

[0153] 1320, network interface;

[0154] 1322, communication link;

[0155] 1324. Communication bus. DETAILED DESCRIPTION

[0156] The following will be combined with the drawings in the embodiments of this specification to clearly and completely describe the technical solutions in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the embodiments of this specification.

[0157] With the continuous development of science and technology and information technology, in the digital age, more and more content, such as advertisements, products, services, activities, etc., need to be recommended and displayed to users. The existing content recommendation method may directly provide a large amount of content to all users without distinction. In this process, some users may be interested in the content products provided, while others may not be interested.

[0158] The existing content recommendation methods are unable to make targeted content recommendations based on the needs and value perceptions of different user groups towards products.

[0159] In order to solve the above problems, an embodiment of this specification provides a content recommendation method. Figure 1 This is a flowchart of a content recommendation method provided by an embodiment of this specification. This specification provides method operation steps as described in the embodiment or flowchart, but may include more or fewer operation steps based on conventional or non-creative labor. The order of steps listed in the embodiment is only one way of executing the order of many steps and does not represent the only execution order. When the actual system or device product is executed, it can be executed in the order of the method shown in the embodiment or the accompanying drawings or in parallel.

[0160] It should be noted that the terms "first", "second", etc. in the description and claims of the embodiments of this specification and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of this specification described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, device, product or equipment that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0161] Reference Figure 1 , this specification embodiment provides a content recommendation method, including:

[0162] S101: Obtain target content corresponding to the user to be recommended and the target content type corresponding to the user to be recommended;

[0163] S102: Divide the target content into modules according to the modules corresponding to the target content type to obtain the modules corresponding to the target content, wherein the modules reflect the information in the target content from different perspectives;

[0164] S103: Calculate the historical preference scores of each module in the target content type corresponding to the user to be recommended and the historical behavior frequencies of all modules in the target content type corresponding to the user to be recommended based on the historical browsing behavior data and historical clicking behavior data of the user to be recommended;

[0165] S104: Calculate the current preference score of each module in the target content type corresponding to the user to be recommended and the current behavior frequency of all modules in the target content type corresponding to the user to be recommended based on the current browsing behavior data and current click behavior data of the user to be recommended;

[0166] S105: Compare the historical behavior frequency and the current behavior frequency, and based on the comparison result, directly or indirectly sort the modules according to the historical preference scores and the current preference scores;

[0167] S106: Obtaining at least one target module corresponding to the to-be-recommended user and its priority according to the direct sorting result or the indirect sorting result;

[0168] S107: configuring at least one target module in the target content based on the priority, and recommending the configured target module to a user to be recommended.

[0169] The step of obtaining target content corresponding to the user to be recommended further includes:

[0170] Determining whether the user to be recommended is a specific user according to the user information of the user to be recommended;

[0171] If yes, determining the target content corresponding to the user to be recommended;

[0172] If not, extracting the user features of the user to be recommended from the associated information of the user to be recommended;

[0173] The user characteristics are input into the trained neural network model to obtain the target content corresponding to the user to be recommended.

[0174] Generally speaking, user information includes user ID, name, nationality, gender, age, etc., and content can be advertisements, products, services, activities, etc. The content recommendation in the embodiment of this specification is to recommend content to the recommended user through a network platform, and the specific media include smart phones, tablet computers or desktop computers, etc.

[0175] Among them, specific users include whitelist users and tagged users. For whitelist users, the whitelist records the user ID. The user ID in the user information of the user to be recommended is obtained and matched with the user ID recorded in the whitelist. If the match is successful, it means that the specific user is a whitelist user. For tagged users, the user information of the user to be recommended is obtained. If there is a pre-set tag in the user information, it means that the user to be recommended is a tagged user.

[0176] For whitelist users and tagged users, the target content to be recommended can be determined directly. For example, if the tagged user is a college student and the target content is advertising, it can be summer card discount advertisements, etc., if the target content is a product, it can be a laptop computer, etc., if the target content is a service, it can be employment consulting services for college students, etc., if the target content is an activity, it can be off-campus practical activities for college students, etc.

[0177] If the user to be recommended is not a specific user, it is necessary to determine the target content corresponding to the user to be recommended. The specific method is to extract the user features of the user to be recommended from the associated information of the user to be recommended, and the associated information includes the user information, device information, account information and consumption information of the user to be recommended. The user features are input into the trained neural network model to obtain the target content corresponding to the user to be recommended. The neural network model can be trained using known historical users. The content of interest of the historical users is known. By obtaining the user features of the historical users and inputting them into the model, the content of interest of the historical users is used as the true value, and the results of the model output are trained to obtain a trained neural network model. The trained neural network model can input the user features of the user to be recommended and output the target content of the user to be recommended.

[0178] Regardless of whether the user to be recommended is a specific user or not, the corresponding target content can be determined. After obtaining the target content, it is necessary to further divide the target content into modules. After the module division, several modules of the target content are obtained. The several modules reflect the information in the target content from several different angles. After the module division, the content is managed in a refined manner to better meet the user's personalized needs, thereby improving the user experience.

[0179] Further, according to the browsing behavior data and click behavior data of the user to be recommended, at least one target module in the target content corresponding to the user to be recommended is determined. The target module represents the information that the user is interested in among all the information in the target content, and at least one target module is configured. When configuring, due to different configuration methods, the target module will be set in different areas of the user's browsing field of view. For example, when the user browses the screen through a smart phone, tablet computer or desktop computer, the field of view includes a focus area and a non-focus area, and the user's attention to the focus area is higher than that of the non-focus area. The area above 1 / 2 of the screen can be set as the focus area, and the area below 1 / 2 of the screen can be set as the non-focus area. The area in the middle of the screen can also be set as the focus area, and the remaining area can be set as the non-focus area. The target modules are configured according to their different priorities, and the ones with higher priorities are set in the focus area. In this way, the configured target modules are recommended to the user to be recommended, which can better meet the needs of the user and improve the user experience.

[0180] Among them, refer to Figure 2 , the extracting the user features of the user to be recommended from the associated information of the user to be recommended further comprises:

[0181] S201: extracting age characteristics, gender characteristics, location characteristics, education level characteristics and occupation characteristics from the user information of the user to be recommended;

[0182] S202: extracting device type features and operating system type features from the device information of the user to be recommended;

[0183] S203: extracting account login time features, account level features, and account payment preference features from the account information of the user to be recommended;

[0184] S204: Extracting consumption type features, consumption frequency features, and consumption amount features from the consumption information of the user to be recommended.

[0185] The location feature in the user information refers to the current location of the user to be recommended, and the device type feature in the device information refers to whether the device used by the user to be recommended is a smartphone, tablet or desktop computer. The account login time feature in the account information refers to the usual login time of the user to be recommended, the account level feature refers to the current account level of the user to be recommended, and the account payment preference feature refers to the usual payment method of the user to be recommended, such as credit card payment, e-wallet payment or other payment software payment. The consumption type feature in the consumption information refers to the type of products that the user to be recommended usually consumes, the consumption frequency feature refers to the frequency of consumption by the user to be recommended, and the consumption amount feature refers to the amount of consumption by the user to be recommended.

[0186] The above features are used as user features of the user to be recommended and input into the trained neural network model to obtain the target content corresponding to the user to be recommended.

[0187] In the embodiment of the present specification, the target content type is at least one of an activity recommendation type, a product recommendation type, an advertisement recommendation type, and a service recommendation type.

[0188] Generally, a content type correspondence table can be set up, which records the contents corresponding to different content types, and the contents corresponding to different content types can be set up according to actual needs. After obtaining the target content, the target content type corresponding to the target content can be determined by searching the content type correspondence table. The modules corresponding to each content type are different, and the target content is divided into modules according to the modules corresponding to the target content type.

[0189] Specifically, dividing the target content into modules according to the modules corresponding to the target content type to obtain the modules corresponding to the target content further includes:

[0190] If the target content type is an activity recommendation type, then according to several modules corresponding to the activity recommendation type, the target content is divided into an activity introduction module, a discount content module and a participation process module;

[0191] If the target content type is a product recommendation type, then according to several modules corresponding to the product recommendation type, the target content is divided into a product introduction module, a discount content module, a user evaluation module and a purchase process module;

[0192] If the target content type is an advertisement recommendation type, the target content is divided into an advertisement introduction module, a discount content module, a user evaluation module, and an action guidance module according to several modules corresponding to the advertisement recommendation type;

[0193] If the target content type is a service recommendation type, the target content is divided into a service introduction module, a preferential content module, a user evaluation module and a participation process module according to several modules corresponding to the loan product type.

[0194] Among them, the modules corresponding to the activity recommendation class are the activity introduction module, the preferential content module and the participation process module; the modules corresponding to the product recommendation class are the product introduction module, the preferential content module, the user evaluation module and the purchase process module; the modules corresponding to the advertising recommendation class are the advertising introduction module, the preferential content module, the user evaluation module and the action guidance module; the modules corresponding to the service recommendation class are the service introduction module, the preferential content module, the user evaluation module and the participation process module.

[0195] By integrating the historical and current behavior data of the recommended user, personalized content modules are recommended to the recommended user and arranged by priority to ensure that the most relevant content is displayed to the user first. This process not only takes into account the long-term interests of the recommended user, but also combines the immediate needs of the recommended user, so that the recommendation results are more in line with the actual preferences of the recommended user.

[0196] Among them, refer to Figure 3 , the calculating of the historical preference scores of each module in the target content type corresponding to the user to be recommended and the historical behavior frequencies of all modules in the target content type corresponding to the user to be recommended according to the historical browsing behavior data and the historical clicking behavior data further includes:

[0197] S501: extracting the historical browsing times, historical average browsing time and historical post-browsing operations of each module in the target content type according to the historical browsing behavior data of the user to be recommended;

[0198] S502: Counting the number of first operations of clicks after the historical browsing in each module, and calculating a first ratio of the first operation number to the historical browsing number in each module;

[0199] S503: extracting the historical click counts and historical post-click operations of each module in the target content type according to the historical click behavior data of the user to be recommended;

[0200] S504: Counting the number of second operations in each module that are effective conversions after the historical click operation, and calculating a second ratio of the second operation number to the historical click number in each module;

[0201] S505: Standardizing the historical browsing times, historical average browsing duration, and first ratio of each module to obtain the processed historical browsing times, processed historical browsing duration, and processed first ratio of each module;

[0202] S506: performing weighted calculation on the processed historical browsing times, processed historical browsing duration and processed first ratio of each module to obtain a historical browsing preference score of each module;

[0203] S507: Standardizing the historical number of clicks and the second ratio of each module to obtain the processed historical number of clicks and the processed second ratio of each module;

[0204] S508: performing weighted calculation on the processed historical click counts and the processed second ratio of each module to obtain a historical click preference score of each module;

[0205] S509: Combining the historical browsing preference scores and historical click preference scores of each module to obtain the historical preference scores of each module;

[0206] S510: Calculate the historical average browsing frequency according to the total number of historical browsing times of all modules;

[0207] S511: Calculate the historical average click frequency according to the total number of historical clicks of all modules;

[0208] S512: Calculate the average of the historical average browsing frequency and the historical average clicking frequency as the historical behavior frequency of all modules in the target content type corresponding to the user to be recommended.

[0209] The past week, past month, past quarter or past six months can be used as the historical period, and the historical browsing behavior data and historical click behavior data of the recommended users can be used as the browsing behavior data and click behavior data of the historical period. Only the historical browsing times, historical average browsing time and historical post-browsing operations of each module corresponding to the target content type are extracted, and there is no need to extract the modules corresponding to other types.

[0210] Taking the past week as an example, the historical number of views is the number of views of each module in the past week. The specific average unit of the historical average browsing time can be set according to the actual situation. For example, it can be the average browsing time of one day, that is, the average browsing time of one day for each module in the past week. The historical post-browsing operation is the next operation after browsing each module in the past week. The next operation may be clicking on the corresponding module or not clicking on the corresponding module.

[0211] The historical number of clicks is the number of clicks on each module in the past week. The historical post-click actions are the next actions after clicking on each module in the past week. The next action may be an effective conversion or an ineffective conversion. An effective conversion means that the goal of the target content has been achieved. For example, if the target content is event promotion, an effective conversion means that the user to be recommended has participated in the corresponding event. If the target content is product recommendation, an effective conversion means that the user to be recommended has purchased the corresponding product, and so on.

[0212] The number of first operations of clicks after historical browsing in each module is counted, and the first ratio of the first operation number to the historical browsing number in each module is calculated. The first ratio represents the ratio of clicks by the recommended users after browsing the corresponding module. The higher this ratio is, the higher the interest of the recommended users in the corresponding module is.

[0213] Similarly, the number of second operations that are effective conversions after historical clicks in each module is counted, and the second ratio of the second operations in each module to the historical clicks is calculated. The second ratio represents the proportion of recommended users who achieve the target content after clicking the corresponding module. The higher this ratio is, the higher the interest of the recommended users in the corresponding module is.

[0214] The historical browsing times, historical average browsing time and first ratio of each module are further standardized to obtain the processed historical browsing times, processed historical browsing time and processed first ratio of each module. The purpose of the standardization is to ensure that each indicator is comparable on the same scale. After the standardization, the processed historical browsing times, processed historical browsing time and processed first ratio of each module are weighted to obtain the historical browsing preference score of each module, wherein the weights of the processed historical browsing times, processed historical browsing time and processed first ratio can be set according to the actual situation, for example, the weights are 30%, 30% and 40% respectively, the processed historical browsing times are multiplied by the corresponding weights, the processed historical browsing time is multiplied by the corresponding weights, and the processed first ratio is multiplied by the corresponding weights, and finally the three are summed to obtain the historical browsing preference score.

[0215] Similarly, the historical number of clicks and the second ratio of each module are standardized to obtain the processed historical number of clicks and the processed second ratio of each module, and the processed historical number of clicks and the processed second ratio of each module are weighted to obtain the historical click preference score of each module.

[0216] The historical browsing preference scores and historical click preference scores of each module are combined to obtain the historical preference scores of each module.

[0217] After counting the total number of historical page views of all modules in the historical period, you can use daily or weekly units, etc. Assuming that it is used as a daily unit, divide the total number of historical page views by the number of days in the historical period to calculate the historical average page view frequency.

[0218] Based on the same method, the historical average click frequency can be calculated, and the average of the historical average browsing frequency and the historical average click frequency can be calculated as the historical behavior frequency of all modules in the target content type corresponding to the recommended user.

[0219] Among them, refer to Figure 4 The standardization of the historical browsing times, the historical average browsing time and the first ratio of each module to obtain the processed historical browsing times, the processed historical browsing time and the processed first ratio of each module further includes:

[0220] S601: Based on the historical browsing times, historical average browsing durations, and first ratios of each module in each module in multiple different historical time periods, multiple historical browsing times, multiple historical average browsing durations, and multiple first ratios of each module are obtained;

[0221] S602: performing normalization processing on the multiple historical browsing times of each module according to the maximum value and the minimum value among the multiple historical browsing times of each module;

[0222] S603: Calculate the average of the multiple standardized historical browsing times as the historical browsing times after each module is processed;

[0223] S604: performing standardization processing on the multiple historical average browsing durations of each module according to the maximum value and the minimum value among the multiple historical average browsing durations of each module;

[0224] S605: Calculate the average of the multiple standardized historical average browsing time as the historical browsing time after each module is processed;

[0225] S606: performing standardization processing on the multiple first ratios of each module according to the maximum value and the minimum value of the multiple first ratios of each module;

[0226] S607: Calculate an average of the first proportions after the multiple standardization processes as the first proportion after the processing of each module in each module;

[0227] Reference Figure 5 The standardization of the historical click counts and the second ratio of each module to obtain the processed historical click counts and the processed second ratio of each module further includes:

[0228] S608: Based on the historical number of clicks and the second ratio of each module in each module in multiple different historical time periods, obtain multiple historical number of clicks and multiple second ratios of each module;

[0229] S609: performing normalization processing on the multiple historical click counts of each module according to the maximum value and the minimum value among the multiple historical click counts of each module;

[0230] S610: Calculate the average of multiple standardized historical click counts as the historical click count after each module is processed;

[0231] S611: performing standardization processing on the multiple second ratios of each module according to the maximum value and the minimum value among the multiple second ratios of each module;

[0232] S612: Calculate an average value of the second ratios after the multiple standardization processes as the second ratio after the processing of each module in each module.

[0233] In the embodiment of this specification, it is necessary to obtain the historical browsing times, historical average browsing durations, and first ratios of each module in multiple different historical time periods. For example, the multiple different historical time periods can be based on a week as a cycle. There are 4 weeks in the past month, and the 4 weeks are 4 different historical time periods. In this way, multiple historical browsing times, multiple historical average browsing durations, and multiple first ratios of each module can be obtained, for example, 4 historical browsing times, 4 historical average browsing durations, and 4 first ratios.

[0234] The multiple historical browsing times of each module are standardized according to the maximum and minimum values ​​among the multiple historical browsing times of each module.

[0235] The formula for standardization can be:

[0236]

[0237] Among them, x′ is the number of historical views after a certain standardization process, x is the number of historical views, and x min is the minimum value among multiple historical page views, x max The maximum value among multiple historical page views.

[0238] In this way, the number of historical page views of each module after multiple standardization processes can be obtained, and the average value is calculated as the number of historical page views of each module after processing.

[0239] Based on the same method, the historical browsing time and the first proportion of each module after processing can be obtained. Whether it is the number of historical browsing times after processing, the historical browsing time after processing or the first proportion after processing, they are all values ​​between [0,1], so it is convenient to perform a unified scale weighted calculation to obtain the historical browsing preference score of each module.

[0240] Based on the same method, the historical number of clicks and the second ratio of each module can be standardized to obtain the historical number of clicks and the processed second ratio of each module, which makes it convenient to perform weighted calculation on a unified scale to obtain the historical click preference score of each module.

[0241] The average of the historical browsing preference scores and the historical click preference scores of each module may be used as the historical preference score of each module.

[0242] In addition to historical browsing behavior data and historical click behavior data, the embodiments of this specification also consider the user's current browsing behavior data and current click behavior data, specifically:

[0243] Reference Figure 6 , the calculation of the current preference score of each module in the target content type corresponding to the user to be recommended and the current behavior frequency of all modules in the target content type corresponding to the user to be recommended further includes:

[0244] S701: extracting the current browsing times, current average browsing time and current post-browsing operations of each module in the target content type according to the current browsing behavior data of the user to be recommended;

[0245] S702: Counting the number of third operations of clicking after the current browsing in each module, and calculating a third ratio of the third operation number to the current browsing number in each module;

[0246] S703: extracting the current click count and current post-click operations of each module in the target content type according to the current click behavior data of the user to be recommended;

[0247] S704: Counting the number of fourth operations in each module in which the current post-click operation is a valid conversion, and calculating a fourth ratio of the fourth operation number in each module to the current click number;

[0248] S705: performing standardization processing on the current browsing times, current average browsing time and third ratio of each module to obtain the processed current browsing times, processed current browsing time and processed third ratio of each module;

[0249] S706: performing weighted calculation on the processed current browsing times, the processed current browsing duration and the processed third ratio of each module to obtain a current browsing preference score of each module;

[0250] S707: performing standardization processing on the current number of clicks and the fourth ratio of each module to obtain the processed current number of clicks and the processed fourth ratio of each module;

[0251] S708: performing weighted calculation on the processed current click counts of each module and the processed fourth ratio to obtain the current click preference score of each module;

[0252] S709: Combining the current browsing preference score and the current clicking preference score of each module to obtain the current preference score of each module;

[0253] S710: Calculate the current average browsing frequency according to the current total number of browsing times of all modules;

[0254] S711: Calculate the current average click frequency according to the current total number of clicks of all modules;

[0255] S712: Calculate the average of the current average browsing frequency and the current average clicking frequency as the current behavior frequency of all modules in the target content type corresponding to the user to be recommended.

[0256] Among them, refer to Figure 7 , the current browsing times, current average browsing time and third ratio of each module are standardized to obtain the processed current browsing times, processed current browsing time and processed third ratio of each module further including:

[0257] S801: Based on the current browsing times, current average browsing durations, and third ratios of each module in each module in multiple different current time periods, multiple current browsing times, multiple current average browsing durations, and multiple third ratios of each module are obtained;

[0258] S802: performing normalization processing on the multiple current browsing times of each module according to the maximum value and the minimum value among the multiple current browsing times of each module;

[0259] S803: Calculate the average of the current browsing times after multiple standardization processes as the current browsing times after each module in each module;

[0260] S804: performing standardization processing on the multiple current average browsing durations of each module according to the maximum value and the minimum value among the multiple current average browsing durations of each module;

[0261] S805: Calculate the average of the current average browsing time after multiple standardization processes as the current browsing time after each module in each module;

[0262] S806: performing standardization processing on the multiple third ratios of each module according to the maximum value and the minimum value among the multiple third ratios of each module;

[0263] S807: Calculate an average value of the third ratios after multiple standardization processes as the third ratio after each module in each module;

[0264] Reference Figure 8 , the current number of clicks and the fourth ratio of each module are standardized to obtain the current number of clicks and the fourth ratio of each module after processing, further comprising:

[0265] S808: Based on the current number of clicks and the fourth ratio of each module in each module in multiple different current time periods, obtain multiple current number of clicks and multiple fourth ratios of each module;

[0266] S809: performing standardization processing on the multiple current click counts of each module according to the maximum value and the minimum value among the multiple current click counts of each module;

[0267] S810: Calculate the average of the current click counts after multiple standardization processes as the current click count after each module in each module;

[0268] S811: performing standardization processing on the multiple fourth ratios of each module according to the maximum value and the minimum value among the multiple fourth ratios of each module;

[0269] S812: Calculate an average value of the fourth ratios after the multiple standardization processes as the fourth ratio after the processing of each module in each module.

[0270] Among them, the current browsing behavior data and current click behavior data of the recommended user can be the browsing behavior data and click behavior data of the current week, current day or current 4 hours. Only the current browsing times, current average browsing time and current post-browsing operations of each module corresponding to the target content type are extracted. There is no need to extract the modules corresponding to other types.

[0271] In the embodiment of this specification, the process of calculating the current preference score of each module in the target content type corresponding to the user to be recommended and the current behavior frequency of all modules in the target content type corresponding to the user to be recommended based on the current browsing behavior data and the current click behavior data is similar to the process of calculating the historical preference score of each module in the target content type corresponding to the user to be recommended and the historical behavior frequency of all modules in the target content type corresponding to the user to be recommended based on the historical browsing behavior data and the historical click behavior data. For the sake of a concise description of the embodiment of this specification, this will not be repeated. Since it takes into account both the long-term interests of the user to be recommended and the immediate needs of the user to be recommended, the recommendation method of the embodiment of this specification is closer to the actual preferences of the user, and the recommendation efficiency and accuracy are higher.

[0272] In the embodiment of the present specification, the comparing the historical behavior frequency and the current behavior frequency, and based on the comparison result, directly or indirectly sorting according to the historical preference score and the current preference score of each module further includes:

[0273] Comparing the historical behavior frequency with the current behavior frequency;

[0274] If the historical behavior frequency and the current behavior frequency are equal, then the modules are directly sorted according to their historical preference scores and current preference scores;

[0275] If the historical behavior frequency and the current behavior frequency are not equal, an indirect sorting is performed based on the historical preference scores and the current preference scores of each module.

[0276] If the historical behavior frequency is equal to the current behavior frequency, it means that the historical behavior and current behavior of the recommended user have equal strength and are equally important. In this case, direct sorting can be performed; otherwise, indirect sorting is performed.

[0277] Wherein, the direct sorting according to the historical preference scores and current preference scores of each module further includes:

[0278] Sort the historical preference scores in descending order, and use the modules corresponding to the top N historical preference scores as the historical preference modules of the user to be recommended;

[0279] The sorting order corresponding to the historical preference module is used as the priority of the historical preference module;

[0280] Sort the current preference scores in descending order, and use the modules corresponding to the top N current preference scores as the current preference modules of the user to be recommended;

[0281] The sorting order corresponding to the current preference module is used as the priority of the current preference module.

[0282] In the embodiments of the specification, N is a positive integer, and the average priority of the historical preference module and the current preference module refers to the average of the priority of the historical preference module and the priority of the current preference module. For example, the sorting order is 1, module A, 2, module C, 3, module B, then the priority of module A is 1, the priority of module C is 2, and the priority of module B is 3.

[0283] Specifically, obtaining at least one target module corresponding to the to-be-recommended user and its priority according to the direct sorting result further includes:

[0284] If the historical preference module and the current preference module of the user to be recommended are exactly the same, the exactly the same module is used as the target module, and the priority of the target module is obtained according to the average priority of the exactly the same historical preference module and the current preference module;

[0285] If the historical preference module and the current preference module of the user to be recommended are partially the same, the modules that are partially the same are used as the target module, and the priority of the target module is obtained according to the average priority of the historical preference modules and the current preference modules that are partially the same;

[0286] If the historical preference module and the current preference module of the user to be recommended are completely different, the current preference module is used as the target module, and the priority of the current preference module is used as the priority of the target module.

[0287] The indirect sorting according to the historical preference scores and current preference scores of each module further includes:

[0288] Determine the historical weight and current weight based on the ratio of historical behavior frequency to current behavior frequency;

[0289] The historical score of each module is obtained by multiplying the historical preference score of each module by the corresponding historical weight;

[0290] The current score of each module is obtained by multiplying the current preference score of each module by the corresponding current weight;

[0291] Sum the historical scores and current scores of each module to get the final score of each module;

[0292] Sort the final scores in descending order.

[0293] The step of obtaining at least one target module and its priority corresponding to the user to be recommended according to the indirect sorting result further includes:

[0294] The module corresponding to the final score of the top M in the ranking is taken as the target module;

[0295] The sorting order corresponding to the target modules is used as the priority of the target modules.

[0296] Wherein, M is a positive integer and can be set according to actual conditions. The historical weight and the current weight can be set according to actual conditions, for example, the historical weight is set to 60% and the current weight is set to 40%, or the historical weight is set to 40% and the current weight is set to 60%, etc. Of course, the historical weight and the current weight can also be determined by other methods. Specifically, the historical weight and the current weight are further determined according to the ratio of the historical behavior frequency to the current behavior frequency. The ratio of the historical weight to the current weight is equal to the ratio of the historical behavior frequency to the current behavior frequency. For example, if the historical behavior frequency is 3 and the current behavior frequency is 9, the ratio is 1:3, the historical weight is 25%, and the current weight is 75%.

[0297] Reference Fig. 9 After obtaining the target module and the priority of the target module, configuring at least one target module in the target content based on the priority, and recommending the configured target module to the user to be recommended further comprises:

[0298] S901: Based on the priorities, a target module with a higher priority in the target content is set in a focus area of ​​the browsing field of the user to be recommended, and a target module with a lower priority is set in a non-focus area of ​​the browsing field of the user to be recommended, to obtain configured target modules;

[0299] S902: Recommend the configured target module to the user to be recommended.

[0300] According to the above description, the user attention of the focus area is higher than that of the non-focus area. The area of ​​the upper 1 / 2 of the screen can be set as the focus area, and the area of ​​the lower 1 / 2 of the screen can be set as the non-focus area. The area of ​​the middle 1 / 2 of the screen can also be set as the focus area, and the remaining area can be set as the non-focus area. For example, the priority of the target modules can be sorted from high to low. If there are an even number of target modules, the 1st to n / 2th target modules are set in the focus area, and the (n / 2)+1th to nth target modules are set in the non-focus area. If there are an odd number of target modules, the 1st to (n+1) / 2th target modules are set in the focus area, and the [(n+1) / 2]+1th to nth target modules are set in the non-focus area, where n is the number of target modules. Finally, the configured target modules are obtained and recommended to the recommended users.

[0301] By judging whether the user to be recommended belongs to a specific user group, if so, the corresponding target content is directly matched. This direct matching is based on accurate identification of specific user needs to ensure that the recommended content is relevant. For non-specific users, user features are extracted from related information, and these features are input into the trained neural network model for prediction to generate personalized target content. This process relies on deep learning technology, which can identify users' potential preferences and needs, thereby ensuring accurate content recommendations.

[0302] Next, the target content is divided into multiple modules through the module division method. These modules reflect the core information of the content from different angles. Based on the user's browsing and clicking behavior data, the modules that the user is interested in are determined and prioritized. The determination of priority depends on the specific performance of the user in the historical behavior data, and the relevant information can be displayed to the user first, further improving the accuracy of the recommendation.

[0303] Configure the recommended content module according to the priority to ensure that the recommended content is in line with the user's interests and can guide the user to generate effective conversions. Flexible adjustment of content through modular configuration to match it with user needs, and maximize the accuracy of content recommendations by combining user information, behavioral data and model predictions.

[0304] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with the relevant laws, regulations and standards of the relevant countries and regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0305] This application provides users with corresponding big data analysis (such as personal biometrics, identity data, consumption data, asset data, electronic terminal operation data, etc.) operation entrances for users to choose to agree or reject automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered.

[0306] Based on the content recommendation method described above, an embodiment of this specification also provides a content recommendation device. The device may include a system (including a distributed system), software (application), module, component, server, client, etc. that uses the method described in the embodiment of this specification and is combined with necessary implementation hardware. Based on the same innovative concept, the device in one or more embodiments provided in the embodiment of this specification is as described in the following embodiments. Since the implementation scheme of the device to solve the problem is similar to the method, the implementation of the specific device in the embodiment of this specification can refer to the implementation of the aforementioned method, and the repeated parts will not be repeated.

[0307] As used below, the term "unit" or "module" may be a combination of software and / or hardware that implements a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0308] Specifically, Fig.10 is a schematic diagram of a module structure of an embodiment of a content recommendation device provided in an embodiment of this specification, referring to Fig.10 As shown, a content recommendation device provided by an embodiment of the present specification includes: an acquisition module 100, a division module 200, a history calculation module 300, a current calculation module 400, a comparison module 500, a target module and priority determination module 600, and a configuration module 700.

[0309] The acquisition module 100 is used to acquire the target content corresponding to the user to be recommended and the target content type corresponding to the user to be recommended;

[0310] A division module 200 is used to divide the target content into modules according to the modules corresponding to the target content type, so as to obtain the modules corresponding to the target content, wherein the modules reflect the information in the target content from different perspectives;

[0311] The history calculation module 300 is used to calculate the historical preference score of each module in the target content type corresponding to the user to be recommended, and the historical behavior frequency of all modules in the target content type corresponding to the user to be recommended, based on the historical browsing behavior data and historical clicking behavior data of the user to be recommended;

[0312] The current calculation module 400 is used to calculate the current preference score of each module in the target content type corresponding to the user to be recommended, and the current behavior frequency of all modules in the target content type corresponding to the user to be recommended, based on the current browsing behavior data and the current clicking behavior data of the user to be recommended;

[0313] A comparison module 500 is used to compare the historical behavior frequency with the current behavior frequency, and based on the comparison result, directly or indirectly sort the modules according to the historical preference scores and the current preference scores;

[0314] The target module and priority determination module 600 is used to obtain at least one target module and its priority corresponding to the user to be recommended according to the direct sorting result or the indirect sorting result;

[0315] The configuration module 700 is used to configure at least one target module in the target content based on the priority, and recommend the configured target module to the user to be recommended.

[0316] Reference Fig.11 As shown, based on the content recommendation method described above, a computer device 1302 is also provided in one embodiment of this specification, wherein the method described above is run on the computer device 1302. The computer device 1302 may include one or more processors 1304, such as one or more central processing units (CPUs) or graphics processing units (GPUs), and each processing unit may implement one or more hardware threads. The computer device 1302 may also include any memory 1306, which is used to store any kind of information such as code, settings, data, etc. In a specific implementation, a computer program on the memory 1306 and executable on the processor 1304, when the computer program is executed by the processor 1304, can execute instructions according to the method described above.

[0317] Without limitation, for example, the memory 1306 may include any one or more combinations of the following: any type of RAM, any type of ROM, flash memory device, hard disk, optical disk, etc. More generally, any memory may use any technology to store information. Further, any memory may provide volatile or non-volatile retention of information. Further, any memory may represent a fixed or removable component of the computer device 1302. In one case, when the processor 1304 executes the associated instructions stored in any memory or combination of memories, the computer device 1302 may perform any operation of the associated instructions. The computer device 1302 also includes one or more drive mechanisms 1308 for interacting with any memory, such as a hard disk drive mechanism, an optical disk drive mechanism, etc.

[0318] The computer device 1302 may also include an input / output module 1310 (I / O) for receiving various inputs (via input devices 1312) and for providing various outputs (via output devices 1314). A specific output mechanism may include a presentation device 1316 and an associated graphical user interface 1318 (GUI). In other embodiments, the input / output module 1310 (I / O), input device 1312, and output device 1314 may not be included, and the computer device 1302 may be used as a computer device in a network. The computer device 1302 may also include one or more network interfaces 1320 for exchanging data with other devices via one or more communication links 1322. One or more communication buses 1324 couple the components described above together.

[0319] The communication link 1322 may be implemented in any manner, for example, through a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication link 1322 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc., governed by any protocol or combination of protocols.

[0320] Corresponds to Figure 1-Figure 11 The method in the embodiment of the present specification also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are executed.

[0321] The embodiment of the present specification also provides a computer-readable instruction, wherein when the processor executes the instruction, the program therein causes the processor to execute the following Figures 1 to 9 The method shown.

[0322] The present specification also provides a computer program product. When the computer program product is executed by a processor of a computer device, the computer program product performs the following steps: Figures 1 to 9 The method shown.

[0323] The computer program product described in this specification is a software product that mainly implements the method described in this specification through a computer program.

[0324] It should be understood that in the various embodiments of this specification, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this specification.

[0325] It should also be understood that in the embodiments of this specification, the term "and / or" is only a description of the association relationship of the associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the embodiments of this specification generally indicates that the associated objects before and after are in an "or" relationship.

[0326] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the embodiments of this specification can be implemented with electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the embodiments of this specification.

[0327] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0328] In the several embodiments provided in this specification, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or it can be an electrical, mechanical or other form of connection.

[0329] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of this specification.

[0330] In addition, each functional unit in each embodiment of this specification may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0331] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0332] Based on this understanding, the technical solution of the embodiments of this specification is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of this specification. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0333] Specific embodiments are used in this specification to illustrate the principles and implementation methods of the embodiments of this specification. The description of the above embodiments is only used to help understand the methods and core ideas of the embodiments of this specification. At the same time, for those skilled in the art, according to the ideas of the embodiments of this specification, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the embodiments of this specification.

Claims

1. A content recommendation method, characterized in that: include: Obtain the target content corresponding to the user to be recommended and the target content type corresponding to the user to be recommended; Divide the target content into modules according to the modules corresponding to the target content type to obtain the modules corresponding to the target content, wherein the modules reflect the information in the target content from different perspectives; According to the historical browsing behavior data and historical clicking behavior data of the user to be recommended, the historical preference score of each module in the target content type corresponding to the user to be recommended and the historical behavior frequency of all modules in the target content type corresponding to the user to be recommended are calculated; According to the current browsing behavior data and current click behavior data of the user to be recommended, the current preference score of each module in the target content type corresponding to the user to be recommended and the current behavior frequency of all modules in the target content type corresponding to the user to be recommended are calculated; Compare the historical behavior frequency with the current behavior frequency, and based on the comparison result, directly or indirectly sort the modules according to the historical preference scores and the current preference scores; Obtaining at least one target module and its priority corresponding to the user to be recommended according to the direct sorting result or the indirect sorting result; At least one target module in the target content is configured based on the priority, and the configured target module is recommended to a user to be recommended.

2. The method according to claim 1, characterized in that The step of obtaining target content corresponding to the user to be recommended further includes: Determining whether the user to be recommended is a specific user according to the user information of the user to be recommended; If yes, determining the target content corresponding to the user to be recommended; If not, extracting the user features of the user to be recommended from the associated information of the user to be recommended; The user characteristics are input into the trained neural network model to obtain the target content corresponding to the user to be recommended.

3. The method according to claim 2, characterized in that The specific users include whitelist users and tagged users.

4. The method according to claim 2, characterized in that: The step of extracting the user characteristics of the user to be recommended from the associated information of the user to be recommended further comprises: Extracting age characteristics, gender characteristics, location characteristics, education level characteristics and occupation characteristics from the user information of the user to be recommended; Extracting device type features and operating system type features from the device information of the user to be recommended; Extracting account login time features, account level features, and account payment preference features from the account information of the user to be recommended; The consumption type characteristics, consumption frequency characteristics and consumption amount characteristics are extracted from the consumption information of the user to be recommended.

5. The method according to claim 1, characterized in that The target content type is at least one of an activity recommendation type, a product recommendation type, an advertisement recommendation type, and a service recommendation type.

6. The method according to claim 5, characterized in that The module division of the target content according to the modules corresponding to the target content type to obtain the modules corresponding to the target content further comprises: If the target content type is an activity recommendation type, then according to several modules corresponding to the activity recommendation type, the target content is divided into an activity introduction module, a discount content module and a participation process module; If the target content type is a product recommendation type, then according to several modules corresponding to the product recommendation type, the target content is divided into a product introduction module, a discount content module, a user evaluation module and a purchase process module; If the target content type is an advertisement recommendation type, the target content is divided into an advertisement introduction module, a discount content module, a user evaluation module, and an action guidance module according to several modules corresponding to the advertisement recommendation type; If the target content type is a service recommendation type, the target content is divided into a service introduction module, a preferential content module, a user evaluation module and a participation process module according to several modules corresponding to the loan product type.

7. The method according to claim 5, characterized in that The step of calculating the historical preference scores of each module in the target content type corresponding to the user to be recommended and the historical behavior frequencies of all modules in the target content type corresponding to the user to be recommended based on the historical browsing behavior data and the historical clicking behavior data of the user to be recommended further comprises: Extract the historical browsing times, historical average browsing time and historical post-browsing operations of each module in the target content type according to the historical browsing behavior data of the user to be recommended; Counting the number of first operations of clicks after the historical browsing in each module, and calculating a first ratio of the first operation number to the historical browsing number in each module; Extracting the historical click counts and historical post-click operations of each module in the target content type according to the historical click behavior data of the user to be recommended; Counting the number of second operations in each module that are effective conversions after the historical click operation, and calculating a second ratio of the second operation number to the historical click number in each module; Standardizing the historical browsing times, historical average browsing duration, and first ratio of each module to obtain the processed historical browsing times, processed historical browsing duration, and processed first ratio of each module; Performing weighted calculation on the processed historical browsing times, processed historical browsing durations and processed first ratios of each module to obtain a historical browsing preference score for each module; Standardize the historical number of clicks and the second ratio of each module to obtain the processed historical number of clicks and the processed second ratio of each module; Performing weighted calculation on the processed historical click counts and the processed second ratio of each module to obtain the historical click preference score of each module; The historical browsing preference scores and historical click preference scores of each module are combined to obtain the historical preference scores of each module; According to the total number of historical views of all modules, the historical average viewing frequency is calculated; According to the total number of historical clicks of all modules, the historical average click frequency is calculated; Calculate the average of the historical average browsing frequency and the historical average click frequency as the historical behavior frequency of all modules in the target content type corresponding to the recommended user.

8. The method according to claim 7, characterized in that The standardization of the historical browsing times, the historical average browsing time and the first ratio of each module to obtain the processed historical browsing times, the processed historical browsing time and the processed first ratio of each module further includes: Based on the historical browsing times, historical average browsing durations, and first ratios of each module in each module in multiple different historical time periods, multiple historical browsing times, multiple historical average browsing durations, and multiple first ratios of each module are obtained; According to the maximum and minimum values ​​of the multiple historical browsing times of each module, the multiple historical browsing times of each module are standardized; Calculate the average of multiple standardized historical browsing times as the historical browsing times after each module is processed; According to the maximum and minimum values ​​of the multiple historical average browsing durations of each module, the multiple historical average browsing durations of each module are standardized; Calculate the average of the historical average browsing time after multiple standardization processes as the historical browsing time after each module is processed; According to the maximum value and the minimum value of the multiple first ratios of each module, the multiple first ratios of each module are standardized; Calculate the average of the first proportions after multiple standardization processes as the first proportion after each module in each module; The standardizing of the historical click counts and the second ratio of each module to obtain the processed historical click counts and the processed second ratio of each module further includes: Based on the historical number of clicks and the second ratio of each module in each module in multiple different historical time periods, multiple historical number of clicks and multiple second ratios of each module are obtained; According to the maximum and minimum values ​​of the multiple historical click counts of each module, the multiple historical click counts of each module are standardized; Calculate the average of multiple standardized historical click counts as the historical click count after each module is processed; According to the maximum value and the minimum value of the multiple second ratios of each module, the multiple second ratios of each module are standardized; An average value of the second ratios after the multiple standardization processes is calculated as the second ratio after the processing of each module in each module.

9. The method according to claim 5, characterized in that The step of calculating the current preference score of each module in the target content type corresponding to the user to be recommended and the current behavior frequency of all modules in the target content type corresponding to the user to be recommended based on the current browsing behavior data and the current click behavior data of the user to be recommended further includes: Extracting the current browsing times, current average browsing time and current post-browsing operations of each module in the target content type according to the current browsing behavior data of the user to be recommended; Counting the number of third operations in each module in which the operation after the current browsing is click, and calculating the third ratio of the number of third operations in each module to the current browsing number; Extracting the current click count and current post-click operations of each module in the target content type according to the current click behavior data of the user to be recommended; Counting the number of fourth operations in each module in which the current post-click operation is a valid conversion, and calculating the fourth ratio of the fourth operation number in each module to the current click number; Standardizing the current browsing times, current average browsing time and third ratio of each module to obtain the processed current browsing times, processed current browsing time and processed third ratio of each module; Performing weighted calculation on the processed current browsing times, processed current browsing duration and processed third ratio of each module to obtain the current browsing preference score of each module; Performing standardization processing on the current number of clicks and the fourth ratio of each module to obtain the processed current number of clicks and the processed fourth ratio of each module; Performing weighted calculation on the processed current click counts and the processed fourth ratio of each module to obtain the current click preference score of each module; The current browsing preference score and the current clicking preference score of each module are integrated to obtain the current preference score of each module; According to the sum of the current browsing times of all modules, the current average browsing frequency is calculated; According to the sum of the current click times of all modules, the current average click frequency is calculated; Calculate the average of the current average browsing frequency and the current average clicking frequency as the current behavior frequency of all modules in the target content type corresponding to the recommended user.

10. The method according to claim 9, characterized in that The standardization of the current browsing times, the current average browsing time and the third ratio of each module to obtain the processed current browsing times, the processed current browsing time and the processed third ratio of each module further includes: Based on the current browsing times, current average browsing durations and third ratios of each module in each module in multiple different current time periods, multiple current browsing times, multiple current average browsing durations and multiple third ratios of each module are obtained; According to the maximum and minimum values ​​of the multiple current browsing times of each module, the multiple current browsing times of each module are standardized; Calculate the average of the current page views after multiple standardization processes as the current page views after each module in each module; According to the maximum and minimum values ​​of the multiple current average browsing durations of each module, the multiple current average browsing durations of each module are standardized; Calculate the average of the current average browsing time after multiple standardization processes as the current browsing time after each module in each module; According to the maximum value and the minimum value of the multiple third ratios of each module, the multiple third ratios of each module are normalized; Calculate the average of the third proportions after multiple standardization processes as the third proportion after each module in each module; The standardization of the current number of clicks and the fourth ratio of each module to obtain the processed current number of clicks and the processed fourth ratio of each module further includes: Based on the current number of clicks and the fourth ratio of each module in each module in multiple different current time periods, multiple current number of clicks and multiple fourth ratios of each module are obtained; According to the maximum and minimum values ​​of the multiple current click times of each module, the multiple current click times of each module are standardized; Calculate the average of the current click counts after multiple standardization processes as the current click count after each module in each module; According to the maximum value and the minimum value of the plurality of fourth ratios of each module, the plurality of fourth ratios of each module are normalized; An average value of the fourth ratios after the multiple standardization processes is calculated as the fourth ratio after the processing of each module in each module.

11. The method according to claim 4, characterized in that The comparing the historical behavior frequency with the current behavior frequency and, based on the comparison result, directly or indirectly sorting according to the historical preference score and the current preference score of each module further includes: Comparing the historical behavior frequency with the current behavior frequency; If the historical behavior frequency and the current behavior frequency are equal, then the modules are directly sorted according to their historical preference scores and current preference scores; If the historical behavior frequency and the current behavior frequency are not equal, an indirect sorting is performed based on the historical preference scores and the current preference scores of each module.

12. The method according to claim 11, characterized in that The direct sorting according to the historical preference scores and current preference scores of each module further includes: Sort the historical preference scores in descending order, and use the modules corresponding to the top N historical preference scores as the historical preference modules of the user to be recommended; The sorting order corresponding to the historical preference module is used as the priority of the historical preference module; Sort the current preference scores in descending order, and use the modules corresponding to the top N current preference scores as the current preference modules of the user to be recommended; The sorting order corresponding to the current preference module is used as the priority of the current preference module.

13. The method according to claim 12, characterized in that The step of obtaining at least one target module and its priority corresponding to the to-be-recommended user according to the direct sorting result further includes: If the historical preference module and the current preference module of the user to be recommended are exactly the same, the exactly the same module is used as the target module, and the priority of the target module is obtained according to the average priority of the exactly the same historical preference module and the current preference module; If the historical preference module and the current preference module of the user to be recommended are partially the same, the modules that are partially the same are used as the target module, and the priority of the target module is obtained according to the average priority of the historical preference modules and the current preference modules that are partially the same; If the historical preference module and the current preference module of the user to be recommended are completely different, the current preference module is used as the target module, and the priority of the current preference module is used as the priority of the target module.

14. The method according to claim 11, characterized in that The indirect sorting according to the historical preference scores and current preference scores of each module further includes: Determine the historical weight and current weight based on the ratio of historical behavior frequency to current behavior frequency; The historical score of each module is obtained by multiplying the historical preference score of each module by the corresponding historical weight; The current score of each module is obtained by multiplying the current preference score of each module by the corresponding current weight; Sum the historical scores and current scores of each module to get the final score of each module; Sort the final scores in descending order.

15. The method according to claim 14, characterized in that The step of obtaining at least one target module and its priority corresponding to the user to be recommended according to the indirect sorting result further includes: The module corresponding to the final score of the top M ranked modules is taken as the target module; The sorting order corresponding to the target modules is used as the priority of the target modules.

16. The method according to claim 1, characterized in that The configuring at least one target module in the target content based on the priority and recommending the configured target module to the user to be recommended further comprises: Based on the priorities, target modules with higher priorities in the target content are set in the focus area of ​​the browsing field of the user to be recommended, and target modules with lower priorities are set in the non-focus area of ​​the browsing field of the user to be recommended, to obtain configured target modules; Recommend the configured target module to the recommended user.

17. A content recommendation device, characterized in that: The device comprises: An acquisition module is used to acquire the target content corresponding to the user to be recommended and the target content type corresponding to the user to be recommended; A division module, used to divide the target content into modules according to the modules corresponding to the target content type, so as to obtain the modules corresponding to the target content, wherein the modules reflect the information in the target content from different perspectives; A history calculation module, used to calculate the historical preference score of each module in the target content type corresponding to the user to be recommended, and the historical behavior frequency of all modules in the target content type corresponding to the user to be recommended, based on the historical browsing behavior data and historical click behavior data of the user to be recommended; A current calculation module, used to calculate the current preference score of each module in the target content type corresponding to the user to be recommended, and the current behavior frequency of all modules in the target content type corresponding to the user to be recommended, based on the current browsing behavior data and current click behavior data of the user to be recommended; A comparison module, used to compare the historical behavior frequency with the current behavior frequency, and based on the comparison result, directly or indirectly sort according to the historical preference score and the current preference score of each module; A target module and priority determination module, used to obtain at least one target module and its priority corresponding to the user to be recommended according to the direct sorting result or the indirect sorting result; A configuration module is used to configure at least one target module in the target content based on the priority, and recommend the configured target module to a user to be recommended.

18. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that: When the computer program is executed by the processor, the computer program executes the instructions of the method according to any one of claims 1 to 16.

19. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor of a computer device, the computer program executes instructions of the method according to any one of claims 1 to 16.

20. A computer program product, characterized in that When the computer program product is executed by a processor of a computer device, the computer program product executes instructions of the method according to any one of claims 1 to 16.

Images