Processing method, processing device and electronic device

Abstract

This application discloses a processing method, processing apparatus, and electronic device. The method, apparatus, and electronic device obtain user information of a target user, determine first recommended content and second recommended content based on the user information of the target user, wherein the first recommended content meets relevant conditions with the target user, and the second recommended content does not meet relevant conditions with the target user, and then perform fusion processing on the determined first recommended content and second recommended content, and recommend the fused content obtained after fusion processing to the target user.

Description

Technical Field

[0001] This application belongs to the field of Internet content recommendation technology, and in particular relates to a processing method, processing device and electronic device. Background Technology

[0002] E-commerce platforms and other websites are increasingly offering more precise content recommendations (such as advertising and product recommendations) to improve browsing efficiency and satisfaction, and increase user frequency. However, more precise content recommendations mean a deeper exposure to user information or characteristics, which contradicts the principles of privacy protection and information security. Furthermore, for privacy-sensitive users, this may actually hinder improvements in satisfaction, usage frequency, and retention. Summary of the Invention

[0003] Therefore, this application discloses the following technical solution:

[0004] A processing method, the method comprising:

[0005] Obtain user information of the target user;

[0006] A first recommended content and a second recommended content are determined based on the user information; the first recommended content meets the relevant conditions with the target user, while the second recommended content does not meet the relevant conditions with the target user;

[0007] The first recommended content and the second recommended content are merged.

[0008] The fused content, obtained after the fusion process, is recommended to the target user.

[0009] Optionally, obtaining the target user's user information includes at least one of the following:

[0010] Obtain the user characteristic information of the target user;

[0011] Obtain the target user's historical operational behavior information in the recommendation system.

[0012] Optionally, determining the first recommended content and the second recommended content based on the user information includes:

[0013] The recommended content is determined to be content whose relevance to the user feature information reaches a first threshold, and / or the historical operation content corresponding to the first behavior feature in the historical operation content set of the target user is determined according to the historical operation behavior information, and is used as the first recommended content.

[0014] The following methods are used to determine the recommended content in the set of content to be recommended, which has a relevance of less than a second threshold to the user feature information, and / or to determine the historical operation content in the set of historical operation content of the target user that corresponds to the second behavior feature based on the historical operation behavior information, and to use it as the second recommended content.

[0015] Wherein, the second threshold is less than the first threshold.

[0016] Optionally, after recommending the fused content obtained through the fusion process to the target user, the method further includes:

[0017] Obtain the target user's operation information on the integrated content;

[0018] In response to the operation indicated by the operation information being a preset operation, the relevance of the content indicated by the operation information to the target user is adjusted.

[0019] Optionally, adjusting the relevance between the content indicated by the operation information and the target user includes:

[0020] If the content indicated by the operation information is at least a part of the second recommended content.

[0021] Improve the relevance of at least some of the content to the target user, so that when recommending content to the target user, the at least some of the content is avoided from being recommended as secondary content to the target user.

[0022] Optionally, the method further includes:

[0023] Obtain the user attributes of the target user;

[0024] In response to the user attribute being the first attribute, the process of determining the first recommended content and the second recommended content based on the user information is triggered;

[0025] In response to the user attribute being a second attribute, a first recommended content is determined based on the user information, and the first recommended content is recommended to the target user;

[0026] The first attribute is used to characterize the corresponding user as a privacy-sensitive user, and the second attribute is used to characterize the corresponding user as a non-privacy-sensitive user.

[0027] Optionally, the process of determining the user attributes of the target user includes:

[0028] The user attributes of the target user are determined based on at least one of the survey information of the target user, the registration information of the target user, and the historical behavior information.

[0029] A processing apparatus, the apparatus comprising:

[0030] The acquisition module is used to obtain user information of the target user;

[0031] The determining module is used to determine a first recommended content and a second recommended content based on the user information; the first recommended content meets relevant conditions with the target user, while the second recommended content does not meet the relevant conditions with the target user;

[0032] The fusion module is used to fuse the first recommended content and the second recommended content.

[0033] The recommendation module is used to recommend the fused content obtained after fusion processing to the target user.

[0034] Optionally, the acquisition module is further configured to acquire the user attributes of the target user;

[0035] The device further includes a response module for:

[0036] In response to the user attribute being the first attribute, the process of determining the first recommended content and the second recommended content based on the user information is triggered;

[0037] In response to the user attribute being a second attribute, a first recommended content is determined based on the user information, and the first recommended content is recommended to the target user;

[0038] The first attribute is used to characterize the corresponding user as a privacy-sensitive user, and the second attribute is used to characterize the corresponding user as a non-privacy-sensitive user.

[0039] An electronic device, comprising:

[0040] Memory, used to store at least one set of computer instructions;

[0041] A processor for implementing the processing method described in any of the preceding descriptions by invoking and executing the instruction set stored in the memory.

[0042] As can be seen from the above scheme, this application discloses a processing method, processing device, and electronic device. The method, device, and electronic device obtain user information of the target user, determine first recommended content and second recommended content based on the user information of the target user, the first recommended content meets relevant conditions with the target user, and the second recommended content does not meet relevant conditions with the target user. Then, the determined first recommended content and second recommended content are fused, and the fused content obtained after fusion is recommended to the target user. Attached Figure Description

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

[0044] Figure 1 This is a flowchart illustrating one of the processing methods provided in this application;

[0045] Figure 2 This is an example of how recommended content is displayed after invalid recommendations are introduced into precisely recommended content.

[0046] Figure 3 This is another flowchart illustrating the processing method provided in this application;

[0047] Figure 4 This is another flowchart illustrating the processing method provided in this application;

[0048] Figure 5 This is a structural diagram of the processing apparatus provided in this application;

[0049] Figure 6 This is a structural diagram of the electronic device provided in this application. Detailed Implementation

[0050] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0051] This application discloses a processing method, processing device, and electronic device, applicable to, but not limited to, internet content recommendation scenarios. It is used to present precisely recommended content to users in a general recommendation manner by introducing invalid recommendations into precise recommendations or similar scenarios, thereby enhancing the security protection of user privacy without losing the advantages of precise recommendations.

[0052] The processing method provided in this application can be implemented as a system function of a recommendation system (in this application, systems with content recommendation functions, such as internet e-commerce platforms and content creation / sharing platforms, are collectively referred to as "recommendation systems"), or it can be implemented as a plugin / mini-program. This plugin / mini-program can be integrated into the recommendation system to support the use of the recommendation processing functions provided by the plugin / mini-program. Alternatively, it can be implemented as a local function of the execution device, where the processing logic of the method is implemented by calling this local function during the runtime of the recommendation system. The execution device can be, but is not limited to, electronic devices in various general-purpose or special-purpose computing environments or configurations, such as personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor devices, etc.

[0053] See Figure 1 The provided information processing method flowchart indicates that the information processing method provided in this application embodiment includes the following processing steps:

[0054] Step 101: Obtain the target user's user information.

[0055] Target users are those users whose content needs to be recommended by recommendation systems such as internet e-commerce platforms and content creation / sharing platforms. Specifically, they can be registered users or unregistered users of the recommendation system.

[0056] Typically, registered or unregistered users of a recommendation system constitute potential target users during their browsing, collection, product transactions, content editing / creation, and content viewing / consumption activities. During this process, the recommendation system can determine whether to recommend content to a user by identifying whether their client-side actions are pre-defined actions that require content recommendations (e.g., accessing a product display homepage upon logging into an e-commerce platform, or searching for a specific product category from its product database by entering search terms). If the user's actions are identified as pre-defined actions, that user becomes a target user for content recommendations, triggering the process of recommending content to that user—the processing flow of the method described in this application.

[0057] When it is necessary to recommend content to target users, the user information of the target users is first obtained as the basis for the recommendation.

[0058] The obtained user information of the target user can be any one or more of the target user's user characteristics information and the target user's historical operation behavior information in the recommendation system.

[0059] The types of user characteristic information can be pre-configured by the recommendation system based on the system's business characteristics. Taking an internet e-commerce platform as an example, user characteristic information can include, but is not limited to, user gender, age group, occupation type, education level, consumption type (e.g., whether they are impulsive or stable consumers, whether they belong to a high-spending group, or whether they prefer high-value consumption), consumption pattern (e.g., whether they prefer loan or non-loan consumption), and purchase frequency characteristics (e.g., low-frequency purchase, high-frequency purchase), among other user characteristics. For content creation / sharing platforms, such as various short video platforms, user characteristic information can include, but is not limited to, user gender, age group, occupation type, education level, preferred content types for creation / viewing, preferred content types for commenting, and preferred time periods for content creation / viewing, etc.

[0060] In practical applications, the specific configuration and collection of user features for the corresponding recommendation system can be used to obtain the target user's corresponding user feature information for that recommendation system. Furthermore, the recommendation system can, but is not limited to, collect and process a series of user information such as gender, age group, consumption type, consumption pattern, or preferred content types for creation / viewing, and preferred content types for commenting, based on user profile modeling. Correspondingly, when obtaining user feature information, the user profile data of the target user provided by the recommendation system can be extracted as the target user's user feature information.

[0061] The target user's historical operational behavior information in the recommendation system includes, but is not limited to: the target user's historical clicks, browsing, favorites, consumption (purchase / add to cart), content creation, content viewing, and other historical operation types in the recommendation system, as well as the operation frequency, operation time / time period, and corresponding operation content (such as browsing, favorite, or purchase / add to cart products, or watching videos), etc., which are all one or more of these, depending on the target user's actual historical operation in the recommendation system.

[0062] The obtained user characteristic information and / or historical operation behavior information of the target users are used as data basis for content recommendation to the target users.

[0063] Step 102: Determine the first and second recommended content based on the target user's user information.

[0064] The applicant found that while some users want the most accurate recommendations in content recommendation scenarios, there are also a large number of privacy-sensitive users who are very concerned about the use of privacy and even fear overly accurate recommendation algorithms. Furthermore, more accurate content recommendations mean a deeper manifestation of user information or user characteristics, which contradicts the concepts of privacy protection and information security.

[0065] Based on this, when recommending content to target users, this application specifically determines two aspects of recommended content, namely, first recommended content and second recommended content, based on the user information of the target users, and uses them to recommend content to the target users. In other words, the content ultimately recommended to the target users includes both the first recommended content and the second recommended content.

[0066] In this system, the first recommended content meets certain criteria with the target user, while the second recommended content does not. By including the first recommended content that meets the criteria in the recommendations delivered to the target user, the advantage of accurate recommendations is preserved, allowing users to quickly and efficiently find and locate content they are truly interested in within the recommendation system (such as online e-commerce platforms), such as videos or products. Simultaneously, by including the second recommended content that does not meet the criteria in the recommendations delivered to the target user, the first recommended content (i.e., accurate recommendation information) is obscured, preventing overly precise recommendations from revealing user information / characteristics excessively, thus protecting user privacy.

[0067] The aforementioned conditions are used to characterize the strong correlation between the recommended content and the target user. Specifically, they can be set as conditions based on preset relevance thresholds or preset user behavior characteristics, so as to support the measurement of the strength of the correlation between the content and the target user through the included relevance thresholds or user behavior characteristics.

[0068] Optionally, in one embodiment, it is possible to determine the content to be recommended in the set of content to be recommended that has a relevance to the user feature information of the target user that reaches a first threshold, and to regard the determined content to be recommended as the first recommended content that meets the relevance condition with the target user.

[0069] The set of content to be recommended refers to a collection of content that can be recommended by the recommendation system. For example, a general (standard) product recommendation list formed by a series of products provided by an e-commerce platform, or a general (standard) video recommendation list formed by a series of short videos provided by a content creation / sharing platform.

[0070] In this implementation, the content characteristics of each piece of content in the set of recommended content can be combined, such as whether it has high cost-effectiveness, whether it is a trendy product, whether it is a high-end product, and its suitable age group, gender, and education level. The correlation between each piece of content and the target user's different dimensions of characteristics can be quantified to obtain the relevance between the content and the target user's different dimensions of characteristics. This relevance is then normalized and weighted to obtain the relevance between the content and the target user's user characteristic information. Based on this, it is determined whether the final relevance reaches the first threshold set in the relevant conditions. If it does, the content is determined to be part of the first recommended content; otherwise, it is not. This process continues until the filtering termination condition corresponding to the first recommended content is met.

[0071] The filtering end criteria for the first recommended content can be set to any of the following, but are not limited to:

[0072] 11) Obtain the first recommended content of the required number of recommended items;

[0073] 12) The processing time reaches the set first duration;

[0074] 13) Complete the traversal of the content in the set of content to be recommended, or the current traversal progress reaches the set number of processed content items in the set of content to be recommended.

[0075] For example, from a general (standard) product recommendation list of an e-commerce platform, select products whose relevance to the target user's profile data reaches a first threshold and use them as the first recommendation content.

[0076] In other embodiments, the historical operation content of the target user can be determined to correspond to the first behavioral feature based on the target user's historical operation behavior information, and the historical operation content corresponding to the first behavioral feature can be regarded as the first recommended content that meets the relevant conditions with the target user.

[0077] The target user's historical activity content set refers to the collection of content corresponding to the target user's historical actions, including but not limited to the collection of content corresponding to a series of historical actions such as clicks, browsing, favorites, purchases / add-to-cart, comments, or content creation. The aforementioned first behavioral characteristic refers to behavioral characteristics that reflect a user's high level of interest in the content, such as the frequency of clicks / browses / purchases reaching a first preset number, or the browsing time reaching a first preset duration, or the content being favorited and placed in a favorited state.

[0078] In this implementation, the strength of the association between a piece of content and the target user can be determined by the user behavior characteristics corresponding to each piece of content in the target user's historical operation content set. If the user behavior characteristics corresponding to the content are the preset first behavior characteristics, then the content is identified as the first recommended content that meets the relevant conditions of the target user; otherwise, it is not the first recommended content.

[0079] For example, based on user browsing time, number of clicks / views / purchases, and other behavioral characteristics, the system can determine the strength of the association between each product in the target user's historical product set (products that have been clicked / viewed / favorited / purchased in the past) and the target user, and identify products in the target user's historical product set that have been browsed for a first preset time and / or have been clicked / viewed / purchased a first preset number of times as the first recommended content.

[0080] In this implementation, if the number of content items in the target user's historical operation content set is small, resulting in an insufficient number of first recommended content items, the corresponding number of content items that meet similar conditions to the already selected first recommended content can be further searched from the content set to be recommended, such as the general product list, to supplement the first recommended content and make up for the required number of first recommended content items.

[0081] The similarity condition mentioned above can refer to the fact that the feature similarity between different contents reaches a set threshold.

[0082] In practical applications, the two methods mentioned above can also be combined to filter the first recommended content.

[0083] Optionally, if the number of first recommended content items selected by any one or both of the above implementation methods exceeds the required number, it is preferable to remove content items with relatively low relevance to the target user's characteristics from the selected first recommended content, and / or remove content items with relatively low browsing / purchase frequency and relatively short browsing time in the target user's history, until the number of remaining first recommended content items does not exceed the required number.

[0084] In addition to the primary recommended content, it is also necessary to identify secondary recommended content that does not meet the relevant conditions for the target users, and to participate in the content recommendation process.

[0085] Similar to the method for determining the first recommended content, there can also be multiple methods for determining the second recommended content. Optionally, in one embodiment, content in the set of content to be recommended that has a relevance of less than a second threshold to the user characteristic information of the target user is determined, and this determined content is regarded as the second recommended content that does not meet the relevance condition with the target user.

[0086] The second threshold is less than the first threshold mentioned above.

[0087] In this implementation, the content features of each piece of content in the set of recommended content can be combined to quantify the association between each piece of content and different dimensions of the target user's features, thus obtaining the relevance between the content and the target user's different dimensions of features. The relevance between the content and the target user's different dimensions of features is then normalized and weighted to obtain the relevance between the content and the target user's user feature information. Unlike the first recommended content, when filtering the second recommended content, it is determined whether the relevance between the content and the target user's user feature information is lower than a set second threshold. If it is lower, the content belongs to the second recommended content that does not meet the relevance condition with the target user; otherwise, it does not belong to the second recommended content. This process continues until the filtering end condition corresponding to the second recommended content is met.

[0088] The filtering termination criteria for the second recommended content can be set to any of the following, but are not limited to:

[0089] 21) Obtain the required number of second recommended items;

[0090] 22) The processing time has reached the second set duration;

[0091] The second duration can be the same as or different from the first duration mentioned above, without restriction.

[0092] 23) Complete the traversal of the content in the set of content to be recommended, or the current traversal progress reaches the set number of processed content items in the set of content to be recommended.

[0093] For example, from the general (standard) product recommendation list of a certain e-commerce platform, select products whose relevance to the user profile of the target user is lower than a second threshold, and use them as the second recommendation content.

[0094] When implementing this application, if the selection of the first and second recommended content is based on determining the relevance between the content in the set of content to be recommended and the user characteristics of the target user, then the selection processes for the first and second recommended content can be merged. That is, the selection of the first and second recommended content can be completed by performing a single traversal of the set of content to be recommended (when traversing each piece of content, it can be determined whether its relevance to the user characteristic information of the target user is higher than a first threshold or lower than a second threshold). This single traversal can be a complete traversal or a partial traversal of all content in the set of content to be recommended, depending on the actual situation.

[0095] In other embodiments, the historical operation content of the target user can be determined based on the target user's historical operation behavior information, and the historical operation content corresponding to the second behavioral feature can be regarded as the second recommended content that does not meet the relevant conditions with the target user.

[0096] In contrast to the first behavioral characteristic mentioned above, the second behavioral characteristic refers to behavioral characteristics that reflect a user's low interest in the content, such as clicking / browsing / purchasing less than the second preset number of times, browsing time less than the second preset duration, or unsave the content.

[0097] Optionally, the second preset number of times is less than the first preset number of times mentioned above, and the second preset duration is less than the first preset duration mentioned above.

[0098] In this implementation, the user behavior characteristics corresponding to each piece of content in the target user's historical operation content set are used to determine the strength of the association between the content and the target user. If the user behavior characteristics corresponding to the content are the preset second behavior characteristics, then the content is identified as the second recommended content that does not meet the relevant conditions with the target user; otherwise, it is not considered as the second recommended content.

[0099] For example, products whose historical browsing time is less than a second preset time and / or whose number of clicks / views / purchases is less than a second preset number are identified as the second recommended content.

[0100] Similarly, in this implementation, if the number of content items in the target user's historical operation content set is small, resulting in an insufficient number of second recommended content items, the corresponding number of content items that meet similar conditions to the already selected second recommended content can be further searched from the content set to be recommended, such as the general product list, to supplement the second recommended content and make up for the required number of second recommended content items.

[0101] In practical applications, the two methods mentioned above can also be combined to filter the second recommended content.

[0102] Optionally, if the number of second recommended content items selected by any one or two of the above implementation methods exceeds the required number of second recommended content items, it is preferable to remove content items that are relatively highly relevant to the target user's characteristic information from the selected second recommended content items, and / or remove content items that the target user has viewed / purchased relatively frequently or for relatively long periods in the past, until the number of remaining second recommended content items does not exceed the required number.

[0103] Step 103: Merge the first and second recommended content.

[0104] The merging of the first and second recommended content can be a simple direct combination of each first and second recommended content, that is, directly appending each second recommended content to the end of each first recommended content, or directly appending each first recommended content to the end of each second recommended content.

[0105] Alternatively, the first and second recommended content can be merged. This can also involve cross-merging the first and second recommended content, that is, scrambling and merging the first and second recommended content together.

[0106] In practical applications, any of the above implementation methods can be selected to achieve the fusion processing of the first and second recommended contents as needed.

[0107] Step 104: Recommend the fused content obtained after fusion processing to the target user.

[0108] After merging the first and second recommended content to obtain the corresponding merged content, the merged content is pushed to the target user's client page for display, thereby achieving content recommendation to the target user.

[0109] Preferably, when merging the first and second recommended content, the resulting merged content should be displayed on the client page in such a way that each screen contains both the first and second recommended content. This ensures that, from the user's perspective, they can view highly accurate recommended content that interests them on each screen, while also including some invalid recommendations. This approach protects user privacy in the display of recommended information on each screen.

[0110] Furthermore, preferably, during the fusion process, different first recommended content should be sorted in descending order of their relevance to the target user, so that the first recommended content with higher relevance to the user can be displayed in a more prominent position on the client page, while the recommendation order of different second recommended content can be disregarded.

[0111] Assuming the first set of recommended content contains 6 products: a1, a2...a6, and the relevance of each product to the target user is in the order: a3 > a2 > a6 > a4 > a5 > a1, and the second set of recommended content contains 6 products: b1, b2...b6, then an example of a recommendation page combining the first and second sets of recommended content would look like this: Figure 2 As shown.

[0112] In summary, the processing method of this embodiment, by determining the first recommended content that meets the relevant conditions for the target user and the second recommended content that does not meet the relevant conditions, and by merging the first and second recommended content to recommend content to the target user, proposes and implements a technical approach of introducing invalid recommendations into accurate recommendations. This allows the accurately recommended content to be presented to the user in a general recommendation manner, which not only helps the user to view highly relevant and interesting content from the recommended content, but also, by introducing and merging invalid recommendations, obscures the highly relevant and interesting content for the user, thereby enhancing the security protection of user privacy and further ensuring user information security. Moreover, it improves the satisfaction of privacy-sensitive users and can also increase their usage frequency and retention.

[0113] Alternatively, in practical applications, a joint privacy computing model based on multiple data holders can be introduced into content recommendation, enabling multiple parties to perform joint calculations without disclosing their own sensitive data, thereby further improving the security of user privacy information.

[0114] Optionally, in one embodiment, see [link to relevant documentation]. Figure 3 The flowchart shown illustrates the processing method. The processing method provided in this application, after recommending the integrated content to the target user, may further include the following processing:

[0115] Step 105: Obtain the target user's operation information on the integrated content.

[0116] After pushing the integrated content to the target user's client, this embodiment detects and obtains the target user's operation information on the integrated content (first recommended content and second recommended content) in real time.

[0117] The target user's interaction information with the integrated content can be, but is not limited to, any one or more actions performed by the target user on the required content within the integrated content, such as clicking, browsing, saving / unsaving, purchasing, commenting, or adding to the blacklist, including click count, browsing duration, and positive / negative sentiment information reflected in the comment information.

[0118] Step 106: In response to the operation indicated by the operation information being a preset operation, adjust the relevance between the content indicated by the operation information of the target user and the target user.

[0119] The preset operation can refer to an operation that indicates a change in the target user's level of interest in the recommended content. This can include, but is not limited to, browsing for a set duration, adding to favorites / unfavorites, leaving a review, purchasing, or adding to a blacklist. For example, if a previously favorited item on the recommendation page is unfavorited, it indicates a decrease in interest. Conversely, if a previously unviewed item on the recommendation page is viewed for more than a set duration or is added to favorites / purchased, it indicates an increase in interest.

[0120] After obtaining the target user's operation information on the integrated content, it is determined whether the operation indicated by the operation information is a preset operation. If it is a preset operation, the relevance of the operation information to the target user is adaptively adjusted based on the target user's actual operation information on the integrated content.

[0121] First, based on the operation information, the system analyzes and identifies the direction of change in the target user's interest in the content indicated by the operation information. For example, if the target user unfavorites a product or blacklists a video, it indicates a decrease in the target user's interest in that product or video. Conversely, if the target user browses a product for more than a set time, favorites it, or makes a purchase, it indicates an increase in the target user's interest in that product. Then, based on the direction of change in the target user's interest in the content indicated by the operation information, the relevance between the content and the target user is adjusted in a consistent direction. Specifically, if the operation information indicates a decrease in the target user's interest in the indicated content, the relevance between the indicated content and the target user is reduced; conversely, if it indicates an increase in interest, the relevance between the indicated content and the target user is increased.

[0122] Specifically, if the operation indicated by the target user's operation information is a preset operation, and the content indicated by the target user's operation information is at least part of the second recommended content, then the relevance of the at least part of the content to the target user is increased, and the increased relevance can prevent the at least part of the content from being recommended to the target user as the second recommended content when making subsequent content recommendations to the target user.

[0123] In other words, if a target user performs one or more of the aforementioned preset operations on at least a portion of the second recommended content, it means that the target user has developed some interest in that portion of the content, and it will no longer be included as an invalid recommendation in subsequent recommendations. Whether that portion of the content is included as the first recommended content in subsequent recommendations depends on the relevance of that portion of the content after adjustment and the algorithm requirements.

[0124] Optionally, in this case, at least a portion of the second recommended content can be added to the target user's historical activity content set. By setting a relatively high relevance value between the content and the target user, or based on the corresponding behavioral characteristics (such as the browsing time exceeding the set time value, or being favorited / purchased), the relevance between the content and the target user can be improved, ensuring that when invalid recommended content (second recommended content) is searched next time, the content will not be selected due to its high relevance to the target user.

[0125] This embodiment can track the target user's attention / interest in the recommended content in real time and dynamically, which facilitates subsequent reasonable recommendations to the target user that are more in line with their dynamic interests.

[0126] Optionally, in one embodiment, see [link to relevant documentation]. Figure 4 The processing method flowchart shown in this application may further include the following processes:

[0127] Step 401: Obtain the user attributes of the target user. If the user attributes of the target user are the first attribute, proceed to step 102. If the user attributes of the target user are the second attribute, proceed to step 402.

[0128] The first attribute is used to characterize the corresponding user as a privacy-sensitive user, and the second attribute is used to characterize the corresponding user as a non-privacy-sensitive user.

[0129] User attributes of the target user can be determined in advance based on at least one of the following: research information on the target user, registration information of the target user, and historical behavior information.

[0130] For example, recommendation systems can collect relevant information about target users in advance through surveys and questionnaires, such as whether they prefer precise recommendations or conventional general recommendations (choosing general recommendations indicates that users are more sensitive to precise recommendations and pay more attention to their privacy protection), whether they allow the extraction of their personal registration information for content recommendations, and which parts of their personal registration information they allow to be extracted for content recommendations, etc. Based on the collected relevant information about target users, the system can analyze their level of privacy sensitivity and thus identify their user attributes; or, based on user registration information, such as personality tags and the completeness of personal private information, the system can analyze the user's level of privacy sensitivity and thus identify their user attributes; or, based on the user's historical operation behavior information, the system can determine whether the user has a large number of invalid operations. If so, it means that the user has a high level of privacy sensitivity and can be identified as a privacy-sensitive user, and their attribute is identified as the first attribute; otherwise, it is identified as the second attribute.

[0131] The aforementioned invalid operations refer to operations that do not result in conversion (such as collection, purchase, etc.) or are not aimed at conversion. Typically, some users may intentionally perform some of the above-mentioned invalid operations to interfere with the user feature recognition of the backend system in order to avoid the recommendation system from capturing their personal information too accurately.

[0132] In other implementations, user attributes can be initially set as the first attribute by default, and users can adjust their attributes as needed. Alternatively, the recommendation system backend can adaptively adjust user attributes based on the user profile data after obtaining sufficient user profile data.

[0133] When it is necessary to recommend content to a target user, this embodiment first obtains the target user's user attributes before determining the recommended content. If the target user's user attribute is the first attribute, then proceed to step 102, based on... Figure 1 The process shown introduces invalid recommendations into the recommended content. This ensures that highly accurate recommendations are pushed to the target user, while also introducing some invalid recommendations into the highly accurate recommendations, thus obfuscating and protecting them. Otherwise, if it is the second attribute, proceed to step 402.

[0134] Step 402: Determine the first recommended content based on the target user's user information, and recommend the determined first recommended content to the target user.

[0135] If the target user's user attribute is the second attribute, meaning the target user is not a privacy-sensitive user, then highly accurate recommended content, i.e., the first recommended content, can be filtered and pushed to the target user to meet the highly accurate recommendation needs of such users.

[0136] When it is necessary to recommend content to target users, there is no restriction on the order of the different steps of obtaining user attributes and user information (user feature information, historical operation behavior information) of target users. Either step can be performed first and the other can be performed later, or they can be performed simultaneously.

[0137] This embodiment uses the privacy-sensitive characteristics represented by user attributes to determine whether to provide precise or inaccurate recommendations to users. This can further achieve content recommendations that match user characteristics, improve the satisfaction of different types of users with content recommendations, and consequently increase the frequency of use and retention of different types of users with the recommendation system.

[0138] It's worth noting that in traditional content recommendation, the recommendation page may also contain content that users are interested in and content that they are not interested in. However, the processing method used by traditional technologies to make these recommendations differs from that in this application. Traditional technologies aim for highly accurate recommendations by filtering content. However, when it is difficult to filter out enough highly accurate recommendations due to insufficient user profile data, insufficient user history records, or insufficient backend processing capabilities, strategies such as random recommendations or general recommendations (e.g., top daily sales) are used. _ To supplement the required number of recommendations (e.g., to fill one screen's worth of items), traditional technology's recommendation pages often rely on random or general recommendation strategies to fill in content that users are not interested in. This means that user interest is highly random, and the backend cannot identify or know whether a user is interested, nor does it address user privacy as a technical issue. This application, however, combines precise recommendation and privacy protection. It filters highly precise content (first recommendation content) and invalid content (second recommendation content). By selectively filtering and integrating these two aspects, the recommendation page includes both highly precise and invalid content, preserving the advantages of highly precise recommendations while protecting user privacy.

[0139] Corresponding to the above processing method, this application also discloses a processing apparatus, the structure of which is as follows: Figure 5 As shown, it includes:

[0140] Module 501 is used to obtain user information of the target user;

[0141] The determining module 502 is used to determine a first recommended content and a second recommended content based on the user information; the first recommended content meets relevant conditions with the target user, while the second recommended content does not meet the relevant conditions with the target user;

[0142] The fusion module 503 is used to fuse the first recommended content and the second recommended content;

[0143] The recommendation module 504 is used to recommend the fused content obtained after fusion processing to the target user.

[0144] In one embodiment, the acquisition module 501 is configured to perform at least one of the following processes:

[0145] Obtain the user characteristic information of the target user;

[0146] Obtain the target user's historical operational behavior information in the recommendation system.

[0147] In one embodiment, the determining module 502 is specifically used for:

[0148] The recommended content is determined to be content whose relevance to the user feature information reaches a first threshold, and / or the historical operation content corresponding to the first behavior feature in the historical operation content set of the target user is determined according to the historical operation behavior information, and is used as the first recommended content.

[0149] The following methods are used to determine the recommended content in the set of content to be recommended, which has a relevance of less than a second threshold to the user feature information, and / or to determine the historical operation content in the set of historical operation content of the target user that corresponds to the second behavior feature based on the historical operation behavior information, and to use it as the second recommended content.

[0150] Wherein, the second threshold is less than the first threshold.

[0151] In one embodiment, the acquisition module 501 is further configured to acquire the target user's operation information on the fused content;

[0152] The above-mentioned device further includes: a response module, used to adjust the relevance between the content indicated by the operation information and the target user in response to the operation indicated by the operation information being a preset operation.

[0153] In one embodiment, when adjusting the relevance between the content indicated by the operation information and the target user, the response module is specifically configured to: if the content indicated by the operation information is at least a portion of the second recommended content, increase the relevance between the at least a portion of the content and the target user, so that when recommending content to the target user, the at least a portion of the content is avoided from being recommended as the second recommended content to the target user.

[0154] In one embodiment, the acquisition module 501 is further configured to acquire the user attributes of the target user;

[0155] The response module is also used for:

[0156] In response to the user attribute being the first attribute, the process of determining the first recommended content and the second recommended content based on the user information is triggered;

[0157] In response to the user attribute being a second attribute, a first recommended content is determined based on the user information, and the first recommended content is recommended to the target user;

[0158] The first attribute is used to characterize the corresponding user as a privacy-sensitive user, and the second attribute is used to characterize the corresponding user as a non-privacy-sensitive user.

[0159] In one embodiment, the user attributes of the target user are determined based on at least one of the survey information of the target user, the registration information of the target user, and historical behavior information.

[0160] The processing apparatus disclosed in this application corresponds to the processing method disclosed in the above method embodiments, so the description is relatively simple. For related similarities, please refer to the description of the above method embodiments, which will not be described in detail here.

[0161] This application also discloses an electronic device, which may be, but is not limited to, a device in a variety of general or special computing device environments or configurations, such as: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor devices, etc.

[0162] The composition and structure of electronic devices, such as Figure 6 As shown, it includes at least:

[0163] Memory 10 is used to store the computer instruction set;

[0164] Computer instruction sets can be implemented in the form of computer programs.

[0165] Processor 20 is configured to implement the processing method disclosed in any of the above method embodiments by executing a computer instruction set.

[0166] The processor 20 can be a central processing unit (CPU), an application-specific integrated circuit (ASIC), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices.

[0167] Electronic devices have a display device and / or have a display interface and can connect to an external display device.

[0168] Optionally, the electronic device may also include a camera assembly, and / or be connected to an external camera assembly.

[0169] In addition to these components, electronic devices may also include communication interfaces, communication buses, and other parts. Memory, processor, and communication interface communicate with each other through the communication bus.

[0170] Communication interfaces are used for communication between electronic devices and other devices. Communication buses can be Peripheral Component Interconnect (PCI) buses or Extended Industry Standard Architecture (EISA) buses, and can be categorized into address buses, data buses, control buses, etc.

[0171] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0172] For ease of description, the above systems or devices are described separately as various modules or units based on their functions. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware components.

[0173] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence or the part that makes a creative contribution, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0174] Finally, it should be noted that in this document, relational terms such as first, second, third, and fourth are used to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0175] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A processing method, the method comprising: Obtain user information of the target user, wherein the user information of the target user includes: obtaining user characteristic information of the target user; Determining first and second recommended content based on the user information includes: identifying recommended content in the set of recommended content whose relevance to the user feature information reaches a first threshold, and using this as the first recommended content; identifying recommended content in the set of recommended content whose relevance to the user feature information is lower than a second threshold, and using this as the second recommended content; wherein the second threshold is less than the first threshold; the first recommended content and the target user satisfy a relevant condition, while the second recommended content and the target user do not satisfy the relevant condition; and performing a fusion process on the first and second recommended content. The fused content obtained after fusion processing is recommended to the target user, wherein the fused content obtained after fusion processing can include first recommended content and second recommended content on each screen when the page is displayed.

2. The method according to claim 1, wherein obtaining the user information of the target user further includes: Obtain the target user's historical operational behavior information in the recommendation system.

3. The method according to claim 2, wherein determining the first recommended content and the second recommended content based on the user information further includes: Based on the historical operation behavior information, the historical operation content of the target user that corresponds to the first behavioral feature is determined and used as the first recommended content. Based on the historical operation behavior information, the historical operation content of the target user that corresponds to the second behavioral feature is determined and used as the second recommended content.

4. The method according to claim 1, further comprising, after recommending the fused content obtained after fusion processing to the target user: Obtain the target user's operation information on the integrated content; In response to the operation indicated by the operation information being a preset operation, the relevance of the content indicated by the operation information to the target user is adjusted.

5. The method according to claim 4, wherein adjusting the relevance between the content indicated by the operation information and the target user includes: If the content indicated by the operation information is at least a part of the second recommended content, the relevance of the at least a part of the content to the target user is improved, so that when recommending content to the target user, the at least a part of the content is avoided as the second recommended content.

6. The method according to claim 1, further comprising: Obtain the user attributes of the target user; In response to the user attribute being the first attribute, the process of determining the first recommended content and the second recommended content based on the user information is triggered; In response to the user attribute being a second attribute, a first recommended content is determined based on the user information, and the first recommended content is recommended to the target user; The first attribute is used to characterize the corresponding user as a privacy-sensitive user, and the second attribute is used to characterize the corresponding user as a non-privacy-sensitive user.

7. The method according to claim 6, wherein the process of determining the user attributes of the target user includes: The user attributes of the target user are determined based on at least one of the survey information of the target user, the registration information of the target user, and the historical behavior information.

8. A processing apparatus, the apparatus comprising: The acquisition module is used to obtain user information of the target user, wherein the user information of the target user includes: obtaining user feature information of the target user; The determining module is configured to determine first recommended content and second recommended content based on the user information, including: determining recommended content in the set of recommended content whose relevance to the user feature information reaches a first threshold, and using it as the first recommended content; determining recommended content in the set of recommended content whose relevance to the user feature information is lower than a second threshold, and using it as the second recommended content; wherein the second threshold is less than the first threshold; the first recommended content and the target user meet a relevant condition, and the second recommended content and the target user do not meet the relevant condition; The fusion module is used to fuse the first recommended content and the second recommended content. The recommendation module is used to recommend the fused content obtained after fusion processing to the target user. The fused content obtained after fusion processing can include first recommended content and second recommended content on each screen when the page is displayed.

9. The apparatus according to claim 8, wherein the acquisition module is further configured to acquire user attributes of the target user; The device further includes a response module for: In response to the user attribute being the first attribute, the process of determining the first recommended content and the second recommended content based on the user information is triggered; In response to the user attribute being a second attribute, a first recommended content is determined based on the user information, and the first recommended content is recommended to the target user; in, The first attribute is used to characterize the corresponding user as a privacy-sensitive user, and the second attribute is used to characterize the corresponding user as a non-privacy-sensitive user.

10. An electronic device, comprising: Memory, used to store at least one set of computer instructions; A processor for implementing the processing method as described in any one of claims 1-7 by invoking and executing the instruction set stored in the memory.

Images