WO2021092803A1 - 推送用户确定方法、装置、服务器以及存储介质 - Google Patents

推送用户确定方法、装置、服务器以及存储介质 Download PDF

Info

Publication number
WO2021092803A1
WO2021092803A1 PCT/CN2019/118100 CN2019118100W WO2021092803A1 WO 2021092803 A1 WO2021092803 A1 WO 2021092803A1 CN 2019118100 W CN2019118100 W CN 2019118100W WO 2021092803 A1 WO2021092803 A1 WO 2021092803A1
Authority
WO
WIPO (PCT)
Prior art keywords
content
pushed
user
word vector
verified
Prior art date
Application number
PCT/CN2019/118100
Other languages
English (en)
French (fr)
Inventor
郭子亮
Original Assignee
深圳市欢太科技有限公司
Oppo广东移动通信有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市欢太科技有限公司, Oppo广东移动通信有限公司 filed Critical 深圳市欢太科技有限公司
Priority to CN201980099734.0A priority Critical patent/CN114303351A/zh
Priority to PCT/CN2019/118100 priority patent/WO2021092803A1/zh
Publication of WO2021092803A1 publication Critical patent/WO2021092803A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor

Definitions

  • This application relates to the field of content push technology, and more specifically, to a method, device, server, and storage medium for determining a push user.
  • the server generally pushes content to the electronic device for display on the electronic device.
  • this application proposes a method, device, server, and storage medium for pushing users to solve the above-mentioned problems.
  • an embodiment of the present application provides a method for determining a push user.
  • the method includes: acquiring first content to be pushed, and extracting a word vector of the first content to be pushed; The second content to be pushed whose similarity of the word vector of the content satisfies the similarity threshold; the first user to be pushed corresponding to the first content to be pushed is determined, and the second content to be pushed corresponding to the second content to be pushed is determined Pushing the user; the first user to be pushed and the second user to be pushed are jointly determined as the target user to be pushed.
  • an embodiment of the present application provides an apparatus for pushing user determination, the apparatus including:
  • the first content to be pushed acquisition module is used to acquire the first content to be pushed and the word vector of the first content to be pushed; the second content to be pushed acquisition module is used to acquire the information related to the first content to be pushed.
  • the second content to be pushed whose similarity of the word vector satisfies the similarity threshold;
  • the user to be pushed determination module is used to determine the first user to be pushed corresponding to the first content to be pushed, and determine the second content to be pushed The second user to be pushed corresponding to the content; a target user to be pushed determination module, configured to jointly determine the first user to be pushed and the second user to be pushed as the target user to be pushed.
  • an embodiment of the present application provides a server, including a memory and a processor, the memory is coupled to the processor, the memory stores instructions, and when the instructions are executed by the processor, the The processor executes the above method.
  • an embodiment of the present application provides a computer-readable storage medium, and the computer-readable storage medium stores program code, and the program code can be invoked by a processor to execute the foregoing method.
  • the push user determination method, device, server, and storage medium acquire the first content to be pushed, and extract the word vector of the first push content, and obtain the similarity with the word vector of the first push content to satisfy the similarity
  • the threshold of the second content to be pushed, the first user to be pushed corresponding to the first content to be pushed is determined, and the second user to be pushed corresponding to the second content to be pushed is determined, and the first user to be pushed and the second user to be pushed are determined
  • the users are jointly determined to be the target users to be pushed, so that by extracting the word vector of the first content to be pushed, the second content to be pushed whose similarity meets the similarity threshold is obtained, and the first user to be pushed corresponding to the first content to be pushed and
  • the second to-be-pushed users corresponding to the second to-be-pushed content are jointly determined as the target to-be-pushed users, so as to expand the user group for content push and improve the content push effect.
  • FIG. 1 shows a schematic diagram of an application environment that can be used in the push user determination method provided by the embodiments of the present application
  • FIG. 2 shows a schematic flowchart of a method for determining a push user according to an embodiment of the present application
  • FIG. 3 shows a schematic flowchart of a method for determining a push user according to another embodiment of the present application
  • FIG. 4 shows a schematic flowchart of an embodiment of step S230 of the push user determination method shown in FIG. 3 of the present application
  • FIG. 5 shows a schematic flowchart of another embodiment of step S230 of the method for determining a push user shown in FIG. 3 of the present application;
  • FIG. 6 shows a schematic flowchart of another embodiment of step S230 of the push user determination method shown in FIG. 3 of the present application;
  • FIG. 7 shows a schematic flowchart of another embodiment of step S230 of the push user determination method shown in FIG. 3 of the present application.
  • FIG. 8 shows a schematic flowchart of step S260 of the push user determination method shown in FIG. 3 of the present application
  • FIG. 9 shows a schematic flowchart of step S261 of the push user determination method shown in FIG. 8 of the present application.
  • FIG. 10 shows a schematic flowchart of an embodiment of step S2611 of the push user determination method shown in FIG. 9 of the present application;
  • FIG. 11 shows a schematic flowchart of another embodiment of step S2611 of the method for determining a push user shown in FIG. 9 of the present application;
  • FIG. 12 shows a schematic flowchart of still another embodiment of step S2611 of the push user determination method shown in FIG. 9 of the present application;
  • FIG. 13 shows a schematic diagram of exponential attenuation provided by an embodiment of the present application.
  • FIG. 14 shows a schematic flowchart of step S262 of the push user determination method shown in FIG. 8 of the present application
  • FIG. 15 shows a schematic flowchart of an embodiment of step S280 of the push user determination method shown in FIG. 7 of the present application
  • FIG. 16 shows a schematic flowchart of another embodiment of step S280 of the push user determination method shown in FIG. 7 of the present application;
  • FIG. 17 shows a block diagram of modules of the device for pushing user determination provided by an embodiment of the present application.
  • FIG. 18 shows a block diagram of a server used to execute the method for determining a push user according to an embodiment of the present application
  • FIG. 19 shows a storage unit for storing or carrying program code for implementing the method for determining a push user according to an embodiment of the present application according to an embodiment of the present application.
  • Operators generally push content to electronic devices to achieve operational goals.
  • the operator will obtain user portraits (user tags) before pushing.
  • user portraits also known as user roles, are an effective tool for delineating target users, contacting user demands and design directions.
  • User portraits have been widely used in various fields. Applications.
  • User portraits usually contain multiple dimensions of portrait features to characterize the appearance of users. For promotional activities, each activity needs to correspond to some user portrait features. All user portrait features targeted by the activity can form an activity portrait.
  • the inventor has discovered through long-term research, and proposed the push user determination method, device, server, and storage medium provided by the embodiments of this application.
  • the similarity is obtained by extracting the word vector of the first content to be pushed to satisfy the similarity.
  • Threshold second content to be pushed, and the first user to be pushed corresponding to the first content to be pushed and the second user to be pushed corresponding to the second content to be pushed are jointly determined as the target user to be pushed, so as to expand the user group for content pushing , Improve the effect of content push.
  • the specific push user determination method will be described in detail in the subsequent embodiments.
  • FIG. 1 shows a schematic diagram of an application environment that can be used in the push user determination method provided by the embodiments of the present application. It includes an operator server 100 and an electronic device 200.
  • the electronic device 200 and the operator server 100 are in communication connection to implement data interaction.
  • the operator server 100 can send push content to the electronic device 200.
  • the electronic device 200 and the operator server 100 can be connected through a data network or a wireless network.
  • the electronic device 200 and the operator server 100 When the electronic device 200 and the operator server 100 are connected through a data network, the electronic device 200 and the operator server 100 can be connected through a 2G network, a 3G network, 4G network, 5G network, etc., when the electronic device 200 and the operator server 100 are connected through a wireless network, the electronic device 200 and the operator server 100 can be connected through a wireless fidelity WiFi network, which is not limited here.
  • the electronic device 200 can be a smart phone, a tablet computer, a wearable electronic device, etc.
  • the operator server 100 can be a traditional server, a cloud server, etc., which is not limited here.
  • FIG. 2 shows a schematic flowchart of a method for determining a push user according to an embodiment of the present application.
  • the push user determination method is used to obtain the second to-be-pushed content whose similarity meets the similarity threshold by extracting the word vector of the first to-be-pushed content, and compare the first to-be-pushed user and the second to-be-pushed content corresponding to the first to-be-pushed content.
  • the second user to be pushed corresponding to the content to be pushed is jointly determined as the target user to be pushed, so as to expand the user group for content pushing and improve the effect of content pushing.
  • the push user determination method is applied to the push user determination apparatus 200 shown in FIG.
  • the server applied in this embodiment may be a traditional server or a cloud server, which is not limited here.
  • the process shown in FIG. 2 will be described in detail below, and the push user determination method may specifically include the following steps:
  • Step S110 Obtain the first content to be pushed, and extract the word vector of the first content to be pushed.
  • the first content to be pushed may include advertisements, games, articles, audio, video, links, etc.
  • the first content to be pushed may include specific content, or may only include subject content, which is not limited here.
  • the first content to be pushed when the first content to be pushed is an article, the first content to be pushed may include the specific content of the article or only the subject content of the article; when the first content to be pushed is an advertisement, the first content to be pushed It may include the specific content of the advertisement, or may only include the advertised product, etc., which is not limited here.
  • the server when the server obtains the first content to be pushed, it can extract the word vector of the first content to be pushed, wherein the word vector of the first content to be pushed can be extracted through word embedding technology.
  • the first content to be pushed can be set with content tags.
  • the first content to be pushed can be set with XX advertising tags, XX game tags, etc., then the server can extract the first content tags through word vector technology. Push the word vector of the content tag of the content, and use the extracted word vector of the content tag of the first content to be pushed as the word vector of the first content to be pushed.
  • Step S120 Obtain the second content to be pushed whose similarity with the word vector of the first content to be pushed meets the similarity threshold.
  • the second content to be pushed when the word vector of the first content to be pushed is extracted, the second content to be pushed may be acquired based on the word vector of the first content to be pushed, where the acquired second content to be pushed is the same as the first content to be pushed.
  • the word vector similarity of the pushed content meets the similarity threshold, where the similarity threshold of the first to-be-pushed content and the second to-be-pushed content meets the similarity threshold, which can be understood as a match between the first to-be-pushed content and the second to-be-pushed content
  • the degree is greater than the matching degree threshold, etc., which is not limited here.
  • the second content to be pushed may include advertisements, games, articles, audio, video, links, etc.
  • the second content to be pushed may include specific content, or may only include subject content, which is not limited here.
  • the first content to be pushed corresponds to the type of the second content to be pushed, that is, when the first content to be pushed is an advertisement, then the second content to be pushed is an advertisement; when the first content to be pushed is an advertisement When it is a game, then the second content to be pushed is a game.
  • the types of the first content to be pushed and the second content to be pushed do not correspond, for example, the first content to be pushed is a game, and the second content to be pushed is an article containing a game, etc., which are not limited here.
  • the first content to be pushed is the corresponding content of a sub-brand of a certain car
  • the word vector of the first content to be pushed can be used to expand the similarity with the sub-brand of the car to other sub-brands that meet the similarity threshold. Identify the car brands of other sub-brands as the second content to be pushed.
  • the first content to be pushed is the corresponding content of a game launched by a certain game manufacturer. Through the word vector of the first content to be pushed, other games whose similarity with a game launched by the game manufacturer meets the similarity threshold can be expanded , Other games whose similarity with the game meets the similarity threshold are determined as the second content to be pushed.
  • Step S130 Determine a first user to be pushed corresponding to the first content to be pushed, and determine a second user to be pushed corresponding to the second content to be pushed.
  • the server may determine the first user to be pushed corresponding to the first content to be pushed, and after obtaining the second content to be pushed, it may determine the same as the second content to be pushed.
  • the corresponding second user to be pushed, where the first user to be pushed and the second user to be pushed may include the same user or different users, which is not limited here.
  • the server may first determine multiple users, and obtain the attribute information of each of the multiple users, and based on the attribute information of each user, determine the user whose attribute information corresponds to the first content to be pushed from the multiple users.
  • the attribute information may include interest tags, age information, gender information, etc., which are not limited here.
  • the server can obtain the interest tag of each of the multiple users and determine the interest tag of each user. Whether the interest tag corresponds to game A, and whether the interest tag of each user corresponds to game B is determined. Among them, when the user's interest tag corresponds to game A, the user whose interest tag corresponds to game A can be determined as the first user to be pushed, and when the user's interest tag corresponds to game B, the interest tag can be corresponding to game B The user of is determined to be the second user to be pushed.
  • Step S140 The first user to be pushed and the second user to be pushed are jointly determined as the target user to be pushed.
  • the first user to be pushed and the second user to be pushed may be Jointly determine the target user to be pushed, and determine the target user to be pushed as the final push user of the first content to be pushed, so as to expand the user group pushed by the first content to be pushed and improve the push effect of the first content to be pushed.
  • the push user determination method obtains the first content to be pushed, extracts the word vector of the first pushed content, and obtains the second to be pushed whose similarity with the word vector of the first pushed content meets the similarity threshold Content, the first user to be pushed corresponding to the first content to be pushed is determined, and the second user to be pushed corresponding to the second content to be pushed is determined, and the first user to be pushed and the second user to be pushed are jointly determined as the target to be pushed Push users, so as to obtain the second content to be pushed whose similarity meets the similarity threshold by extracting the word vector of the first content to be pushed, and map the first user to be pushed corresponding to the first content to be pushed to the second content to be pushed
  • the second users to be pushed are jointly determined as the target users to be pushed in order to expand the user group for content pushing and improve the effect of content pushing.
  • FIG. 3 shows a schematic flowchart of a method for determining a push user according to another embodiment of the present application.
  • the method is applied to the server, and the process shown in FIG. 3 will be described in detail below.
  • the method for determining the push user may specifically include the following steps:
  • Step S210 Obtain the first content to be pushed, and determine keywords in the first content to be pushed.
  • the first content to be pushed can be identified and keywords in the first content to be pushed can be extracted.
  • the number of keywords extracted from the first content to be pushed can be one or more.
  • the keyword can be directly extracted. Determined as keywords in the first content to be pushed; when the number of keywords extracted from the first content to be pushed is multiple, multiple keywords can be determined as keywords in the first content to be pushed , Or when the number of keywords extracted from the first content to be pushed is multiple, multiple keywords can be filtered, the more important target keywords are filtered from multiple keywords, and the target keywords are determined It is the keyword in the first content to be pushed.
  • Step S220 Convert the keyword into the word vector.
  • the determined keywords may be converted into word vectors, where the word vector obtained by the conversion may be used as the word vector of the first content to be pushed.
  • Step S230 Calculate the word vector similarity of the first content to be pushed and multiple content to be verified respectively.
  • the server may calculate the similarity of the word vectors of the first content to be pushed and multiple content to be verified based on the word vector of the first content to be pushed.
  • multiple content to be verified may be content in a pre-created resource pool.
  • multiple content to be verified may include advertisements, games, articles, audio, video, links, etc., which are not described here. limited.
  • multiple content to be verified can be content in the resource pool that is more popular within a preset time period, for example, it can be content in the resource pool whose conversion rate during the preset time period is higher than a specified conversion rate.
  • the conversion rate is higher than the specified conversion rate can include: the ratio of downloads and views is higher than the first specified ratio, the ratio of installs and views is higher than the second specified ratio, and the ratio of jumps and views is high
  • the third designated ratio, etc. is not limited here.
  • multiple content to be verified can be content in the resource pool whose reference value is higher than the specified reference value, where the reference value is higher than the specified reference value may include: the total number of users is higher than the specified number of users, the total number of comments The number of comments higher than the specified number, the total score is higher than the specified total score, the average score is higher than the specified average score, etc., are not limited here.
  • FIG. 4 shows a schematic flowchart of an embodiment of step S230 of the push user determination method shown in FIG. 3 of the present application.
  • the process shown in FIG. 4 will be described in detail below, and the method may specifically include the following steps:
  • Step S231A extract the word vector of each content to be verified in the plurality of content to be verified respectively.
  • the word vector of each content to be verified in the plurality of content to be verified may be extracted respectively.
  • word embedding technology can be used to extract the word vector of each content to be verified in a plurality of content to be verified.
  • the keywords in each content to be verified can be determined separately, and the keywords in each content to be verified can be converted into word vectors.
  • Step S232A Based on the word vector of the first content to be pushed and the word vector of each content to be verified, respectively calculate the similarity of the word vectors of the first content to be pushed and the plurality of content to be verified.
  • the first content to be pushed may be calculated based on the word vector of the first content to be pushed and the word vector of each content to be verified The word vector similarity with each of the multiple content to be verified.
  • FIG. 5 shows a schematic flowchart of another embodiment of step S230 of the push user determination method shown in FIG. 3 of the present application.
  • the process shown in FIG. 5 will be described in detail below, and the method may specifically include the following steps:
  • Step S231B Input the word vector of the first content to be pushed into a preset word vector model, and the preset word vector model is generated from the word vectors of the multiple content to be verified.
  • the server may input the word vector of the first content to be pushed into the word vector model, where the word vector model is composed of multiple words of the content to be verified.
  • Vector generation Specifically, first collect a training data set, where the attributes or features of one type of data in the training data set are distinguished from another type of data, and then calculate the word vector similarity of the collected training data set, so as to be based on the training data set. Sum up the rules and get the word vector model.
  • one type of data of the training data set is multiple word vectors of the content to be verified and word vectors referring to the pushed content as independent variables
  • the other type of data is multiple of the content to be verified and reference to the pushed content.
  • the word vector similarity calculates the word vector similarity between the word vectors of multiple content to be verified and the word vector of the reference push content, and the correlation between the multiple content to be verified and the reference push content can be obtained, and then Obtain the word vector model.
  • Step S232B Obtain the word vector similarity between the first content to be pushed and the multiple content to be verified outputted by the preset word vector model.
  • the word vector model after the word vector model receives the word vector of the first content to be pushed input by the server, it can process the word vector of the first content to be pushed, and output the first content to be pushed and multiple pieces of content to be verified. For each word vector similarity to be verified and pushed in the content, correspondingly, the server may obtain the word vector similarity of the first to-be-pushed content output by the word vector model and multiple content to be verified.
  • FIG. 6 shows a schematic flowchart of another embodiment of step S230 of the method for determining a push user shown in FIG. 3 of the present application.
  • the process shown in FIG. 6 will be described in detail below, and the method may specifically include the following steps:
  • Step S231C Obtain the subject content of the first content to be pushed.
  • the server may analyze the first content to be pushed to obtain the subject content of the first content to be pushed.
  • the server may perform keyword extraction on the first content to be pushed, and determine the subject content of the first content to be pushed based on the extracted keywords. For example, when the extracted keyword is "XX game", it may It is determined that the subject content of the first content to be pushed is a game; when the extracted keyword is "XX movie", it can be determined that the subject content of the first content to be pushed is a movie.
  • the server may perform word segmentation processing on the first content to be pushed, and determine the number of occurrences of each phrase after word segmentation processing, obtain the phrase with the most occurrences, and determine the first to-be-to-be based on the phrase with the most occurrences.
  • the subject content of the pushed content for example, when the phrase with the most occurrences is "XX game", it can be determined that the subject content of the first content to be pushed is a game; when the phrase with the most occurrences is "XX movie”, it can be determined The subject content of the first content to be pushed is a movie.
  • Step S232C Obtain multiple content to be verified that are the same as the subject content of the first content to be pushed.
  • the subject content of the first content to be pushed may be searched for from all the content included in the resource pool based on the subject content of the first content to be pushed, which is the same as the subject content of the first content to be pushed. And determine the content found as the content to be verified. For example, if the subject content of the first content to be pushed is a game, then the subject content can be queried from all the content included in the resource pool as game-related content, and the found subject content is determined to be game-related content to be verified content.
  • Step S233C Calculate the word vector similarity of the first content to be pushed and the multiple content to be verified respectively.
  • FIG. 7 shows a schematic flowchart of another embodiment of step S230 of the push user determination method shown in FIG. 3 of the present application.
  • the process shown in FIG. 7 will be described in detail below, and the method may specifically include the following steps:
  • Step S231D Obtain the type of the first content to be pushed.
  • the server may analyze the first content to be pushed to obtain the type of the first content to be pushed.
  • the server may pre-establish the mapping relationship between the pushed content and the type, and generate a mapping relationship table and store it in the server, as shown in Table 1.
  • the mapping relationship between the pushed content and the type can be manually associated by the user, can be automatically associated by the server, etc., which is not limited here, and the mapping relationship between the pushed content and the type can include one push content corresponding to one type, and more One push content corresponds to one type, one push content corresponds to multiple types, etc., which are not limited here.
  • the server may compare the first content to be pushed with a plurality of pushed content pre-stored in the mapping relationship table one by one to obtain the pushed content that matches the first content to be pushed, and then According to the mapping relationship table, the type corresponding to the pushed content is searched, so that the type of the first content to be pushed can be obtained. For example, assuming that the first content to be pushed is the first application to be pushed, and the content to be pushed is the push application, then, when the mapping relationship table may include the mapping relationship between WeChat and instant messaging, that is, the mapping relationship in the mapping relationship table WeChat is stored under the push application, and the instant messaging category is stored under the type.
  • WeChat is associated with the instant messaging category. Then, when the first application to be pushed is WeChat, the first application to be pushed can be stored in the mapping relationship table. The push application is matched. It can be understood that the first application to be pushed can be matched with WeChat in the mapping relationship table, and it can be found that the type of the first application to be pushed is an instant messaging type.
  • Step S232D Obtain multiple content to be verified that are of the same type as the first content to be pushed.
  • all content included in the resource pool may be searched for the same type of content as the first content to be pushed. And determine the content found as the content to be verified. For example, if the type of the first content to be pushed is a game type, then the content of the game type can be queried from all contents included in the resource pool, and the found content of the game type is determined as the content to be verified.
  • Step S233D Calculate the word vector similarity of the first content to be pushed and the multiple content to be verified respectively.
  • Step S240 Obtain, from the plurality of content to be verified, the target content to be verified whose word vector similarity to the first content to be pushed is greater than the specified similarity.
  • the server may preset and store a designated similarity, and the designated similarity is used as a basis for judging the similarity between each content to be verified and the word vector of the first content to be pushed.
  • the similarity of the word vector between the content to be verified and the first content to be pushed is greater than the specified similarity, the similarity of the word vector representing the content to be verified and the first content to be pushed is higher, that is, the content to be verified and the first content to be pushed are characterized by a higher similarity.
  • the pushed content is relatively similar, and the content to be verified can be used as the second content to be pushed.
  • the word vector similarity between the content to be verified and the first content to be pushed is not greater than the specified similarity, the similarity of the word vector representing the content to be verified and the first content to be pushed is low, that is, the content to be verified and the first content to be pushed are characterized The content is not similar, and the content to be verified cannot be used as the second content to be pushed.
  • the word vectors of the first content to be pushed and each content to be verified can be respectively similar to the specified word vector
  • the degree of comparison is performed to separately determine whether the word vector similarity between the multiple content to be verified and the first content to be pushed is greater than the specified similarity, and the judgment result is obtained, and based on the judgment result, it is obtained from the multiple content to be verified and the first content to be pushed
  • the content to be verified whose word vector similarity is greater than the specified similarity is determined as the target to-be-verified content whose word vector similarity to the first content to be pushed is greater than the specified similarity.
  • Step S250 Determine the target content to be verified as the second content to be pushed.
  • Step S260 Determine a first user to be pushed corresponding to the first content to be pushed, and determine a second user to be pushed corresponding to the second content to be pushed.
  • step S260 For the specific description of step S260, please refer to step S130, which will not be repeated here.
  • FIG. 8 shows a schematic flowchart of step S260 of the method for determining a push user shown in FIG. 3 of the present application.
  • the process shown in FIG. 8 will be described in detail below, and the method may specifically include the following steps:
  • Step S261 Obtain current interest tags of multiple users for multiple content, where the multiple content includes the first content to be pushed.
  • the server may obtain user portraits of multiple users in advance and store them locally.
  • the user portrait can be used to characterize the user's interest tag, payment behavior, push conversion rate, etc., for example, the user portrait can be used to characterize the target user's preferred game type, preferred video type, purchasing power, and click-through rate of the pushed content , The download rate of the pushed content, etc.
  • the server may determine, based on the user profile of the user, the user’s current interest tag for multiple content including the first content to be pushed, where the interest tag can be used to characterize the user’s interest in the multiple content, where the interest level can be Including very interested, very interested, relatively interested, or not interested, etc., which are not limited here. For example, it is possible to obtain the user's degree of interest in the first content to be pushed, and determine whether the user is very interested, very interested, more interested, or not interested in the first content to be pushed.
  • FIG. 9 shows a schematic flowchart of step S261 of the push user determination method shown in FIG. 8 of the present application.
  • the process shown in FIG. 9 will be described in detail below.
  • the method may specifically include the following steps:
  • Step S2611 Obtain historical interest tags of the plurality of users for the plurality of contents, and obtain the time attenuation factor of the plurality of users for the plurality of contents.
  • the server may separately record the historical behaviors of multiple users on the multiple contents, and determine the historical interest tags of the multiple users on the multiple contents based on the historical behaviors of the multiple users on the multiple contents.
  • the server can separately record changes in the interest of multiple users in multiple content, and determine the time attenuation factor of multiple users for multiple content based on the changes in interest of multiple users in multiple content, wherein the user's interest in content changes It may include that the user is more and more interested in the content, less and less interested in the content, interested in a preset time period, etc., which is not limited here.
  • FIG. 10 shows a schematic flowchart of an embodiment of step S2611 of the push user determination method shown in FIG. 9 of the present application.
  • the following will elaborate on the process shown in FIG. 10, and the method may specifically include the following steps:
  • Step S26111A Obtain the browsing behavior of the multiple content by the multiple users within the first preset time.
  • the user's browsing behavior on the content can reflect the user's degree of interest in the content. Specifically, the more the user's browsing behavior on the content, the more interested the user is in the content, the less the user's browsing behavior on the content, and the less the user is interested in the content.
  • the content browsing behavior of the user within the first preset time can be acquired, for example, the number of times the user browses the content recorded by the server within the first preset time can be acquired.
  • the first preset time is the first preset duration before the current moment
  • the obtained content browsing behavior of the user within the first preset time may be used to characterize the user's first preset duration before the current moment.
  • the level of interest in the content may be one day, two days, etc., and the first preset time may be automatically set by the server or manually set by the user, which is not limited here.
  • Step S26112A Determine the time attenuation factor of the multiple users for the multiple contents based on the browsing behavior.
  • the time attenuation factor of the user's content may be determined based on the browsing behavior.
  • the browsing behavior may be the number of browsing times
  • the time attenuation factor of the user to the content may be determined based on the first preset time and the number of browsing times.
  • FIG. 11 shows a schematic flowchart of another embodiment of step S2611 of the push user determination method shown in FIG. 9 of the present application.
  • the process shown in FIG. 11 will be described in detail below, and the method may specifically include the following steps:
  • Step S26111B Obtain the download behavior of the multiple contents by the multiple users within a second preset time.
  • the user's download behavior of the content can reflect the user's degree of interest in the content. Specifically, the more the user downloads the content, the more interested the user is in the content, the less the user downloads the content, and the less the user is interested in the content.
  • the content download behavior of the user within the second preset time can be acquired. For example, the number of downloads of the content recorded by the server within the second preset time, such as 0 times, 1 time, etc., can be acquired.
  • the second preset time is the second preset duration before the current moment, and the acquired download behavior of the content by the user within the second preset time can be used to characterize the user's second preset duration before the current moment.
  • the level of interest in the content may be one day, two days, etc., and the second preset time may be automatically set by the server or manually set by the user, which is not limited here.
  • Step S26112B Determine the time attenuation factors of the multiple users for the multiple contents based on the download behavior.
  • the time attenuation factor of the user's content can be determined based on the downloading behavior.
  • the download behavior may be the number of downloads
  • the user's time attenuation factor of the content may be determined based on the second preset time and the number of downloads.
  • FIG. 12 shows a schematic flowchart of still another embodiment of step S2611 of the method for determining a push user shown in FIG. 9 of the present application.
  • the following will elaborate on the process shown in FIG. 12, and the method may specifically include the following steps:
  • Step S26111C Obtain the jump behaviors of the multiple users to the multiple content within a third preset time.
  • the user's jumping behavior to the content can reflect the user's degree of interest in the content. Specifically, the more the user jumps to the content, the more the user is interested in the content, the less the user jumps to the content, the less the user is interested in the content, and the jump behavior can represent the user's interest in the content.
  • the content jump behavior of the user within the third preset time can be acquired, for example, the number of times the user jumps to the content recorded by the server within the third preset time can be acquired.
  • the third preset time is the third preset duration before the current moment, and the obtained content jump behavior of the user within the third preset time may be used to characterize the third preset duration before the current moment.
  • the level of interest in the content may be one day, two days, etc., and the third preset time may be automatically set by the server or manually set by the user, which is not limited here.
  • Step S26112C Determine the time attenuation factors of the multiple users for the multiple contents based on the jump behavior.
  • the time attenuation factor of the user to the content may be determined based on the jump behavior.
  • the download behavior may be the number of jumps
  • the time attenuation factor of the user to the content may be determined based on the third preset time and the number of jumps.
  • Step S2612 Based on the historical interest tags and the time attenuation factor, obtain current interest tags of the multiple users for the multiple contents.
  • the server may determine the user's current interest tag of the content based on the user's historical interest tag and time decay factor of the content. For example, if the historical interest tag of the user in the content indicates that the user is more interested in the content, and the time attenuation factor is decreasing, then the current interest tag of the user in the content may change from being more interested to not interested, that is, it is determined that the user is more interested in the content. The current interest tag of the content is not interested.
  • the ranking of the user's interest tag of the content can be obtained.
  • the ranking of the user's interest label of the content can be obtained by Newton's law of cooling.
  • User interest tags based on the time decay factor can retain tags that users have been interested in, while sudden hot tags that cannot represent the user's true interest will eventually gradually decay over time.
  • Step S262 Determine from the multiple users a target user whose current interest tag meets a preset condition as the first user to be pushed.
  • the server may set and store preset conditions in advance, and the preset conditions are used as a basis for the user to determine the current interest tag of the content.
  • the preset conditions are used as a basis for the user to determine the current interest tag of the content.
  • the interest tag satisfies the preset condition, it indicates that the user has a high current interest in the first content to be pushed, that is, the user can be the first user to be pushed.
  • the interest tag does not meet the preset condition, it indicates that the user's current interest in the first content to be pushed is low, that is, the user cannot be the first user to be pushed.
  • the current interest tags can be compared with preset conditions respectively to determine whether the current interest tags meet the preset conditions.
  • the determination result represents When the current interest tag meets the preset condition, the user corresponding to the current interest tag can be determined, and the user can be determined as the first user to be pushed.
  • FIG. 14 shows a schematic flowchart of step S262 of the push user determination method shown in FIG. 8 of the present application.
  • the process shown in FIG. 14 will be described in detail below, and the method may specifically include the following steps:
  • Step S2621 Obtain the current popularity value of the target user for the first content to be pushed.
  • the current popularity value of the target user for the first content to be pushed can be obtained.
  • the higher the popularity value the more interested the target user is in the first content to be pushed, and the lower the popularity value, the less interested the target user is in the first content to be pushed.
  • Step S2622 When the current popularity value is greater than a specified popularity value, determine the target user as the first user to be pushed.
  • the server may preset and store a specified popularity value, and the specified popularity value may be used as a basis for determining the current popularity value of the first content to be pushed by the target user. Therefore, in this embodiment, after obtaining the current popularity value of the first content to be pushed by the target user, the current popularity value can be compared with the specified popularity value to determine whether the current popularity value is greater than the specified popularity value. Wherein, when the judgment result indicates that the current popularity value is greater than the specified popularity value, it indicates that the target user is currently interested in the first content to be pushed, and the target user may be determined as the first user to be pushed.
  • Step S270 The first user to be pushed and the second user to be pushed are jointly determined as the target user to be pushed.
  • step S270 please refer to step S140, which will not be repeated here.
  • Step S280 Push the first content to be pushed to the target user to be pushed.
  • the server may push the first content to be pushed to the target user to be pushed, so as to expand the user group to which the first content to be pushed is pushed.
  • FIG. 15 shows a schematic flowchart of an embodiment of step S280 of the method for determining a push user shown in FIG. 7 of the present application.
  • the process shown in FIG. 15 will be described in detail below, and the method may specifically include the following steps:
  • Step S281A Obtain a first user portrait of the target user to be pushed, where the first user portrait is used to characterize the display preference of the target user to be pushed with respect to the displayed content.
  • the server may pre-acquire the first user portrait of the target user to be pushed and store it locally. After determining the target user to be pushed, the server may read the first user portrait of the target user to be pushed locally.
  • the first user portrait is used to characterize the display preference of the target user to be pushed for the content displayed on the electronic device.
  • the user portrait may be used to characterize the display position of the target user to be pushed for the content displayed on the electronic device. Preference, preference for display size of displayed content, preference for display color of displayed content, preference for display duration of displayed content, etc., are not limited here.
  • Step S282A Determine a content display mode corresponding to the first user portrait based on the first user portrait of the target user to be pushed.
  • the server may determine the content display mode corresponding to the first user portrait of the target user to be pushed according to the first user portrait of the target user to be pushed. It is understandable that the content display mode determined according to the first user portrait of the target user to be pushed corresponds to the display preference of the target user to be pushed for the content displayed by the electronic device. For example, if the first user portrait represents that the target user’s preference for the display position of the content displayed by the electronic device is the upper left of the electronic device, then the content display mode determined according to the first user portrait of the target user to be pushed corresponds to The placement of is at the top left of the electronic device.
  • a mapping relationship between user portraits and content display modes may be established in advance, and the mapping relationship may include multiple user portraits and multiple content display modes, where multiple user portraits and multiple content display modes can correspond one-to-one , Multiple user portraits can correspond to one content display mode, and one user portrait can correspond to multiple content display modes, etc.
  • the first user portrait of the target user to be pushed can be obtained, and then the first user portrait can be compared with multiple user portraits in the pre-established mapping relationship to obtain a user portrait matching the first user portrait Then, based on the mapping relationship, the content display mode corresponding to the user portrait matching the first user portrait is obtained, and the obtained content display mode is determined as the content display mode corresponding to the first user portrait.
  • Step S283A Determine a push format of the first content to be pushed based on the content display mode, and push the first content to be pushed to the target user to be pushed according to the push format.
  • the push format of the first content to be pushed may be determined based on the determined content display mode, and the push format is sent to the target
  • the user to be pushed pushes the first content to be pushed, so that the final display mode of the first content to be pushed on the electronic device is more in line with the user's preferences, and the conversion rate and user experience of the pushed content can be improved.
  • the push content may be adjusted based on the content display mode, so that the format of the adjusted push content is consistent with the content display mode. For example, if the font size of the pushed content is font size 4, and the determined font size of the content display method is font size 5, then the font size of the pushed content can be adjusted from font size 4 to font size 5 to make the adjustment
  • the font size of the pushed content is the same as the font size of the content display method.
  • FIG. 16 shows a schematic flowchart of another embodiment of step S280 of the push user determination method shown in FIG. 7 of the present application.
  • the process shown in FIG. 16 will be described in detail below, and the method may specifically include the following steps:
  • Step S281B Obtain a second user portrait of the target user to be pushed.
  • the server may pre-acquire the second user portrait of the target user to be pushed and store it locally. After determining the target user to be pushed, the server may read the second user portrait of the target user to be pushed locally.
  • Step S282B Determine the restricted content restricting the target user to be pushed according to the second user portrait of the target user to be pushed.
  • the server may determine the restricted content that restricts the target user to be pushed according to the second user portrait of the target user to be pushed.
  • a mapping relationship between user portraits and restricted content may be established in advance. As shown in Table 2, the mapping relationship may include multiple user portraits and multiple restricted contents, where multiple user portraits and multiple restricted contents may be One-to-one correspondence, multiple user portraits can correspond to one restricted content, one user portrait can correspond to multiple restricted content, etc.
  • the second user portrait of the target user to be pushed can be obtained, and then the second user portrait of the target user to be pushed is compared with multiple user portraits in the pre-established mapping relationship to obtain the target user to be pushed
  • the user portrait matched with the second user portrait of the user, and then based on the mapping relationship, the restricted content corresponding to the user portrait matching the second user portrait of the target user to be pushed is obtained, and the user who matches the second user portrait of the target user to be pushed is determined.
  • the restricted content corresponds to the portrait, the restricted content can be determined as restricted content that restricts the target user to be pushed.
  • the second user portrait includes user attributes.
  • the restriction content for restricting the target user to be pushed can be determined according to the user attribute of the target user to be pushed.
  • user attributes may include user age, user location, user experience, user gender, and so on.
  • the restricted content for restricting the target user to be pushed may be determined according to the user age of the target user to be pushed.
  • the restricted content that restricts the target user to be pushed may be determined according to the user location of the target user to be pushed.
  • the restriction content that restricts the target user to be pushed can be determined according to the user experience of the target user to be pushed.
  • the restricted content that restricts the target user to be pushed may be determined according to the user gender of the target user to be pushed.
  • the restricted content that restricts the target user to be pushed can include takeaway.
  • Push Taking user attributes including user experience as an example, different user experiences correspond to different restricted content that needs to be restricted. For example, if user experience does not include stock trading experience, the restricted content restricting target users to be pushed may include stock trading push.
  • different user ages correspond to different restrictions that need to be restricted. For example, if the user is too young (underage), the restricted content to restrict the target users to be pushed can include game push .
  • Step S283B When the restricted content is not included in the first content to be pushed, push the first content to be pushed to the target user to be pushed.
  • the restricted content after determining the restricted content that restricts the target user to be pushed, it can be determined whether the first content to be pushed includes restricted content. For example, the restricted content can be compared with the first content to be pushed to determine Whether the first content to be pushed includes restricted content.
  • the first content to be pushed can be pushed to the target user to be pushed, so as to expand the user group to which the first content to be pushed is pushed.
  • Step S284B When the first content to be pushed includes the restricted content, it is prohibited to push the first content to be pushed to the target user to be pushed.
  • the judgment result indicates that the first content to be pushed includes restricted content
  • it can be prohibited to push the first content to be pushed to the target user to be pushed to avoid pushing inappropriate content to the target user to be pushed, thereby affecting the user experience and Issues affecting push conversion rate.
  • the method for determining a push user obtains the first content to be pushed, determines the keywords in the first content to be pushed, converts the keywords into word vectors, and calculates the first content to be pushed and multiple The word vector similarity of the content to be verified, the target content to be verified whose word vector similarity to the first content to be pushed is greater than the specified similarity is obtained from multiple content to be verified, and the target content to be verified is determined as the second content to be pushed .
  • this embodiment determines that the similarity with the first content to be pushed satisfies the similarity by calculating the word vector similarity of the first content to be pushed and the content to be verified. Threshold second content to be pushed to improve the accuracy of similar content acquisition.
  • FIG. 17 shows a block diagram of a device 300 for pushing a user according to an embodiment of the present application.
  • the pushing user determining device 300 is applied to the above-mentioned server. The following will describe the block diagram shown in FIG. 17.
  • the pushing user determining device 300 includes: a first to-be-pushed content acquisition module 310, a second to-be-pushed content acquisition module 320, The to-be-pushed user determination module 330 and the target-to-be-pushed user determination module 340, wherein:
  • the first content to be pushed acquisition module 310 is configured to acquire the first content to be pushed, and extract the word vector of the first content to be pushed.
  • the first content to be pushed acquisition module 310 includes: a word vector similarity calculation submodule, a target content to be verified acquisition submodule, and a second content to be pushed determination submodule, wherein:
  • the word vector similarity calculation sub-module is used to calculate the word vector similarity of the first content to be pushed and multiple content to be verified respectively.
  • the word vector similarity calculation sub-module includes: a word vector extraction unit and a first word vector similarity calculation unit, wherein:
  • the word vector extraction unit is used to extract the word vector of each content to be verified in the plurality of content to be verified respectively.
  • the first word vector similarity calculation unit is configured to calculate the first content to be pushed and the plurality of content to be verified based on the word vector of the first content to be pushed and the word vector of each content to be verified The word vector similarity of the content.
  • the word vector similarity calculation sub-module includes: a word vector input unit and a second word vector similarity calculation unit, wherein:
  • the word vector input unit is configured to input the word vector of the first content to be pushed into a preset word vector model, and the preset word vector model is generated from the word vectors of the plurality of content to be verified.
  • the second word vector similarity calculation unit is configured to obtain the word vector similarity of the first content to be pushed and the multiple content to be verified outputted by the preset word vector model.
  • the word vector similarity calculation sub-module includes: a subject content acquisition unit, a first to-be-verified content acquisition unit, and a third word vector similarity calculation unit, wherein:
  • the subject content acquiring unit is configured to acquire the subject content of the first content to be pushed.
  • the first content to be verified acquisition unit is configured to acquire multiple content to be verified that are the same as the subject content of the first content to be pushed.
  • the third word vector similarity calculation unit is configured to calculate the word vector similarity of the first content to be pushed and the multiple content to be verified respectively.
  • the word vector similarity calculation sub-module includes: a type acquisition unit, a second to-be-verified content acquisition unit, and a fourth word vector similarity calculation unit, wherein:
  • the type obtaining unit is configured to obtain the type of the first content to be pushed.
  • the second content-to-be-verified acquiring unit is configured to acquire multiple content-to-be-verified of the same type as the first content to be pushed.
  • the fourth word vector similarity calculation unit is configured to calculate the word vector similarity of the first content to be pushed and the multiple content to be verified respectively.
  • the target to-be-verified content obtaining sub-module is configured to obtain, from the plurality of to-be-verified content, the target to-be-verified content whose word vector similarity to the first content to be pushed is greater than a specified similarity.
  • the second content to be pushed determination sub-module is configured to determine the target content to be verified as the second content to be pushed.
  • the first content to be pushed acquisition module 310 includes: a keyword determination sub-module and a word vector conversion sub-module, wherein:
  • the keyword determination submodule is used to obtain the first content to be pushed and determine the keywords in the first content to be pushed.
  • the word vector conversion sub-module is used to convert the keyword into the word vector.
  • the second content to be pushed acquisition module 320 is configured to acquire the second content to be pushed whose similarity with the word vector of the first content to be pushed meets the similarity threshold.
  • the user to be pushed determination module 330 is configured to determine a first user to be pushed corresponding to the first content to be pushed, and to determine a second user to be pushed corresponding to the second content to be pushed.
  • the user to be pushed determination module 330 includes: a current interest tag acquisition sub-module and a first user to be pushed determination sub-module, wherein:
  • the current interest tag acquisition submodule is used to acquire current interest tags of multiple users for multiple content, and the multiple content includes the first content to be pushed.
  • the current interest tag acquiring submodule includes: a time attenuation factor acquiring unit and a current interest tag acquiring unit, wherein:
  • the time attenuation factor obtaining unit is configured to obtain the historical interest tags of the multiple users for the multiple contents, and obtain the time attenuation factors of the multiple users on the multiple contents.
  • time attenuation factor obtaining unit includes: a browsing behavior obtaining subunit and a first time attenuation factor determining subunit, wherein:
  • the browsing behavior obtaining subunit is configured to obtain browsing behaviors of the plurality of contents of the plurality of users within a first preset time.
  • the first time attenuation factor determination subunit is configured to determine the time attenuation factors of the multiple users for the multiple contents based on the browsing behavior.
  • time attenuation factor obtaining unit includes: a downloading behavior obtaining subunit and a second time attenuation factor determining subunit, wherein:
  • the download behavior obtaining subunit is configured to obtain download behaviors of the plurality of contents by the plurality of users within a second preset time.
  • the second time attenuation factor determination subunit is configured to determine the time attenuation factors of the multiple users for the multiple contents based on the download behavior.
  • time attenuation factor obtaining unit includes: a jump behavior obtaining subunit and a third time attenuation factor determining subunit, wherein:
  • the jumping behavior obtaining subunit is configured to obtain the jumping behaviors of the multiple users to the multiple contents within the third preset time.
  • the third time attenuation factor determination subunit is configured to determine the time attenuation factors of the multiple users for the multiple contents based on the jump behavior.
  • the current interest tag obtaining unit is configured to obtain current interest tags of the multiple users for the multiple contents based on the historical interest tags and the time attenuation factor.
  • the first user to be pushed determination sub-module is configured to determine, from the multiple users, a target user whose current interest tag meets a preset condition as the first user to be pushed.
  • the first user to be pushed determination sub-module includes: a current popularity value acquisition unit and a first user to be pushed determination unit, wherein:
  • the current popularity value obtaining unit is configured to obtain the current popularity value of the target user for the first content to be pushed.
  • the first user to be pushed determination unit is configured to determine the target user as the first user to be pushed when the current popularity value is greater than a specified popularity value.
  • the target user to be pushed determination module 340 is configured to jointly determine the first user to be pushed and the second user to be pushed as the target user to be pushed.
  • the device 300 for determining pushing users further includes: a first content pushing module to be pushed, wherein:
  • the first content to be pushed push module is configured to push the first content to be pushed to the target user to be pushed.
  • the first content pushing module to be pushed includes:
  • the first user portrait obtaining sub-module is configured to obtain a first user portrait of the target user to be pushed, and the first user portrait is used to characterize the display preference of the target user to be pushed with respect to the displayed content.
  • the content display mode determination submodule is configured to determine the content display mode corresponding to the first user portrait based on the first user portrait of the target user to be pushed.
  • the push format determination submodule is configured to determine the push format of the first content to be pushed based on the content display mode, and push the first content to be pushed to the target user to be pushed according to the push format.
  • the first to-be-pushed content pushing module includes: a second user portrait acquisition sub-module, a restricted content determination sub-module, a first to-be-pushed content pushing sub-module, and a first to-be-pushed content prohibiting sub-module, wherein:
  • the second user portrait obtaining sub-module is used to obtain the second user portrait of the target user to be pushed.
  • the restricted content determining sub-module is configured to determine the restricted content for restricting the target user to be pushed according to the second user portrait of the target user to be pushed.
  • the first content to be pushed push submodule is configured to push the first content to be pushed to the target user to be pushed when the restricted content is not included in the first content to be pushed.
  • the first content to be pushed prohibition submodule is configured to prohibit pushing the first content to be pushed to the target user to be pushed when the restricted content is included in the first content to be pushed.
  • the coupling between the modules may be electrical, mechanical or other forms of coupling.
  • each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist alone physically, or two or more modules may be integrated into one module.
  • the above-mentioned integrated modules can be implemented in the form of hardware or software function modules.
  • FIG. 18 shows a structural block diagram of a server 100 provided by an embodiment of the present application.
  • the server 100 in this application may include one or more of the following components: a processor 110, a memory 120, and one or more application programs, where one or more application programs may be stored in the memory 120 and configured to be operated by one or Multiple processors 110 execute, and one or more programs are configured to execute the method described in the foregoing method embodiment.
  • the processor 110 may include one or more processing cores.
  • the processor 110 uses various interfaces and lines to connect various parts of the entire server 100, and executes the server by running or executing instructions, programs, code sets, or instruction sets stored in the memory 120, and calling data stored in the memory 120. 100's various functions and processing data.
  • the processor 110 may adopt at least one of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA).
  • DSP Digital Signal Processing
  • FPGA Field-Programmable Gate Array
  • PDA Programmable Logic Array
  • the processor 110 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), a modem, and the like.
  • the CPU mainly processes the operating system, user interface, and application programs; the GPU is used to render and draw the content to be displayed; the modem is used to process wireless communication. It can be understood that the above-mentioned modem may not be integrated into the processor 110, but may be implemented by a communication chip alone.
  • the memory 120 may include random access memory (RAM) or read-only memory (Read-Only Memory).
  • the memory 120 may be used to store instructions, programs, codes, code sets or instruction sets.
  • the memory 120 may include a program storage area and a data storage area, where the program storage area may store instructions for implementing the operating system and instructions for implementing at least one function (such as touch function, sound playback function, image playback function, etc.) , Instructions used to implement the following various method embodiments, etc.
  • the data storage area can also store data (such as phone book, audio and video data, chat record data) created by the terminal 100 during use.
  • FIG. 19 shows a structural block diagram of a computer-readable storage medium provided by an embodiment of the present application.
  • the computer-readable medium 400 stores program code, and the program code can be invoked by a processor to execute the method described in the foregoing method embodiment.
  • the computer-readable storage medium 400 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read Only Memory), EPROM, hard disk, or ROM.
  • the computer-readable storage medium 400 includes a non-transitory computer-readable storage medium.
  • the computer-readable storage medium 400 has a storage space for executing the program code 410 of any method step in the above-mentioned method. These program codes can be read from or written into one or more computer program products.
  • the program code 410 may be compressed in a suitable form, for example.
  • the push user determination method, device, server, and storage medium provided in the embodiments of the application obtain the first content to be pushed, and extract the word vector of the first push content, and obtain the difference between the word vector of the first push content and The second content to be pushed whose similarity meets the similarity threshold, the first user to be pushed corresponding to the first content to be pushed is determined, and the second user to be pushed corresponding to the second content to be pushed is determined, and the first user to be pushed is determined Determine the target user to be pushed together with the second user to be pushed, so as to obtain the second content to be pushed whose similarity meets the similarity threshold by extracting the word vector of the first content to be pushed, and compare the first content to be pushed corresponding to the first content to be pushed.
  • a user to be pushed and a second user to be pushed corresponding to the second content to be pushed are jointly determined as the target user to be pushed, so as to expand the user group for content pushing and improve the effect of content pushing.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

本申请公开了一种推送用户确定方法、装置、服务器以及存储介质。获取第一待推送内容，并提取第一推送内容的词向量，获取与第一推送内容的词向量的相似度满足相似度阈值的第二待推送内容，确定与第一待推送内容对应的第一待推送用户，并确定与第二待推送内容对应的第二待推送用户，将第一待推送用户和第二待推送用户共同确定为目标待推送用户。本申请实施例提供的推送用户确定方法、装置、服务器以及存储介质通过提取第一待推送内容的词向量来获取相似度满足相似度阈值的第二待推送内容，并将第一待推送内容对应的第一待推送用户和第二待推送内容对应的第二待推送用户共同确定为目标待推送用户，以扩大内容推送的用户群体，提升内容推送效果。

Description

推送用户确定方法、装置、服务器以及存储介质 技术领域
本申请涉及内容推送技术领域，更具体地，涉及一种推送用户确定方法、装置、服务器以及存储介质。
背景技术
随着科学技术的发展，电子设备的使用越来越广泛，功能越来越多，已经成为人们日常生活中的必备之一。目前，服务器一般会向电子设备推送内容，以在电子设备进行展示。
发明内容
鉴于上述问题，本申请提出了一种推送用户确定方法、装置、服务器以及存储介质，以解决上述问题。
第一方面，本申请实施例提供了一种推送用户确定方法，所述方法包括：获取第一待推送内容，并提取所述第一待推送内容的词向量；获取与所述第一待推送内容的词向量的相似度满足相似度阈值的第二待推送内容；确定与所述第一待推送内容对应的第一待推送用户，并确定与所述第二待推送内容对应的第二待推送用户；将所述第一待推送用户和所述第二待推送用户共同确定为目标待推送用户。
第二方面，本申请实施例提供了一种推送用户确定装置，所述装置包括：
第一待推送内容获取模块，用于获取第一待推送内容，并提取所述第一待推送内容的词向量；第二待推送内容获取模块，用于获取与所述第一待推送内容的词向量的相似度满足相似度阈值的第二待推送内容；待推送用户确定模块，用于确定与所述第一待推送内容对应的第一待推送用户，并确定与所述第二待推送内容对应的第二待推送用户；目标待推送用户确定模块，用于将所述第一待推送用户和所述第二待推送用户共同确定为目标待推送用户。
第三方面，本申请实施例提供了一种服务器，包括存储器和处理器，所述存储器耦接到所述处理器，所述存储器存储指令，当所述指令由所述处理器执行时所述处理器执行上述方法。
第四方面，本申请实施例提供了一种计算机可读取存储介质，所述计算机可读取存储介质中存储有程序代码，所述程序代码可被处理器调用执行上述方法。
本申请实施例提供的推送用户确定方法、装置、服务器以及存储介质，获取第一待推送内容，并提取第一推送内容的词向量，获取与第一推送内容的词向量的相似度满足相似度阈值的第二待推送内容，确定与第一待推送内容对应的第一待推送用户，并确定与第二待推送内容对应的第二待推送用户，将第一待推送用户和第二待推送用户共同确定为目标待推送用户，从而通过提取第一待推送内容的词向量来获取相似度满足相似度阈值的第二待推送内容，并将第一待推送内容对应的第一待推送用户和第二待推送内容对应的第二待推送用户共同确定为目标待推送用户，以扩大内容推送的用户群体，提升内容推送效果。
附图说明
为了更清楚地说明本申请实施例中的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其它的附图。
图1示出了可用于本申请实施例提供的推送用户确定方法的应用环境示意图；
图2示出了本申请一个实施例提供的推送用户确定方法的流程示意图；
图3示出了本申请又一个实施例提供的推送用户确定方法的流程示意图；
图4示出了本申请的图3所示的推送用户确定方法的步骤S230的一个实施例的流程示意图；
图5示出了本申请的图3所示的推送用户确定方法的步骤S230的又一个实施例的流程示意图；
图6示出了本申请的图3所示的推送用户确定方法的步骤S230的再一个实施例的流程示意图；
图7示出了本申请的图3所示的推送用户确定方法的步骤S230的另一个实施例的流程示意图；
图8示出了本申请的图3所示的推送用户确定方法的步骤S260的流程示意图；
图9示出了本申请的图8所示的推送用户确定方法的步骤S261的流程示意图；
图10示出了本申请的图9所示的推送用户确定方法的步骤S2611的一个实施例的流程示意图；
图11示出了本申请的图9所示的推送用户确定方法的步骤S2611的又一个实施例的流程示意图；
图12示出了本申请的图9所示的推送用户确定方法的步骤S2611的再一个实施例的流程示意图；
图13示出了本申请实施例提供的指数衰减的示意图；
图14示出了本申请的图8所示的推送用户确定方法的步骤S262的流程示意图；
图15示出了本申请的图7所示的推送用户确定方法的步骤S280的一个实施例的流程示意图；
图16示出了本申请的图7所示的推送用户确定方法的步骤S280的又一个实施例的流程示意图；
图17示出了本申请实施例提供的推送用户确定装置的模块框图；
图18示出了本申请实施例用于执行根据本申请实施例的推送用户确定方法的服务器的框图；
图19示出了本申请实施例的用于保存或者携带实现根据本申请实施例的推送用户确定方法的程序代码的存储单元。
具体实施方式
为了使本技术领域的人员更好地理解本申请方案，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述。
运营方一般会通过向电子设备推送内容以达到运营目的。通常，运营方在推送前会获取用户画像(用户标签)，其中，用户画像又称用户角色，是一种勾画目标用户、联系用户诉求及设计方向的有效工具，用户画像在各领域得到了广泛的应用。用户画像通常包含多个维度的画像特征来表征用户的样子，对于促销活动，每个活动都需要对应一些用户画像特征，活动针对的所有用户画像特征即可形成活动画像，通过活动画像与用户画像进行对比，将包含活动画像的画像特征的用户画像圈进活动人群，即可生成促销活动的活动人群，以针对该活动人群进行活动投放，以实现目标用户的精准圈取，再进行统一、批量的推送，以提升推送内容的点击率与资源利用率，同时最小化无用推送信息对用户体验的负面影响。由于现有的内容资讯的用户群体推送，主要是通过资讯的标题，内容等给每篇资讯打上相应内容标签，再结合用户标签，通过选择和用户标签匹配来获取相应的用户群体。然而，当推送的内容标签较少时，所圈取的人群就会变得很少，推送范围也会变得很窄，推送效果不佳。
针对上述问题，发明人经过长期的研究发现，并提出了本申请实施例提供的推送用户确定方法、装置、服务器以及存储介质，通过提取第一待推送内容的词向量来获取相似度满足相似度阈值的第二待推送内容，并将第一待推送内容对应的第一待推送用户和第二待推送内容对应的第二待推送用户共同确定为目标待推送用户，以扩大内容推送的用户群体，提升内容推送效果。其中，具体的推送用户确定方法在后续的实施例中进行详细的说明。
下面将针对可用于本申请实施例提供的推送用户确定方法的环境示意图进行描述。
请参阅图1，图1示出了可用于本申请实施例提供的推送用户确定方法的应用环境示意图。其包括运营商服务器100和电子设备200，电子设备200和运营商服务器100通信连接以实现数据交互，例如，运营商服务器100可以发送推送内容至电子设备200。其中，电子设备200和运营商服务器100可以通过数据网络或无线网络连接，当电子设备200和运营商服务器100通过数据网络连接时，电子设备200和运营商服务器100可以通过2G网络、3G网络、4G网络、5G网络等连接，当电子设备200和运营商服务器100通过无线网络连接时，电子设备200和运营商服务器100可以通过无线保真WiFi网络连接，在此不做限定。其中，电子设备200可以为智能手机、平板电脑、穿戴式电子设备等，运营商服务器100可以为传统服务器，云服务器等，在此不做限定。
请参阅图2，图2示出了本申请一个实施例提供的推送用户确定方法的流程示意图。所述推送用户确定方法用于通过提取第一待推送内容的词向量来获取相似度满足相似度阈值的第二待推送内容，并将第一待推送内容对应的第一待推送用户和第二待推送内容对应的第二待推送用户共同确定为目标待推送用户，以扩大内容推送的用户群体，提升内容推送效果。在具体的实施例中，所述推送用户确定方法应用于如图17所示的推送用户确定装置200以及配置有推送用户确定装置200的服务器100(图18)。下面将以服务器为例，说明本实施例的具体流程，当然，可以理解的，本实施例所应用的服务器可以为传统服务器，也可以为云服务器，在此不做限定。下面将针对图2所示的流程进行详细的阐述，所述推送用户确定方法具体可以包括以下步骤：
步骤S110：获取第一待推送内容，并提取所述第一待推送内容的词向量。
在一些实施方式中，第一待推送内容可以包括广告、游戏、文章、音频、视频、链接等，另外，第一待推送内容可以包括具体内容，也可以仅包括主题内容，在此不做限定。例如，当第一待推送内容为文章时，该第一待推送内容可以包括文章的具体内容，也可以仅包括文章的主题内容；当第一待推送内容为广告时，该第一待推送内容可以包括广告的具体内容，也可以仅包括广告的产品等，在此不做限定。
在本实施例中，服务器在获取到第一待推送内容时，可以提取该第一待推送内容的词向量，其中，可以通过词向量(word embedding)技术提取第一待推送内容的词向量。在一些实施方式中，第一待推送内容可以设定有内容标签，例如，第一待推送内容可以设定有XX广告标签、XX游戏标签等，那么，服务器可以通过词向量技术提取第一待推送内容的内容标签的词向量，并将提取到的第一待推送内容的内容标签的词向量作为第一待推送内容的词向量。
步骤S120：获取与所述第一待推送内容的词向量的相似度满足相似度阈值的第二待推送内容。
在一些实施方式中，在提取到第一待推送内容的词向量时，可以基于第一待推送内容的词向量获取第二待推送内容，其中，所获取的第二待推送内容与第一待推送内容的词向量相似度满足相似度阈值，其中，第一待推送内容和第二待推送内容的相似度阈值满足相似度阈值，可以理解为第一待推送内容和第二待推送内容的匹配度大于匹配度阈值等，在此不做限定。其中，第二待推送内容可以包括广告、游戏、文章、音频、视频、链接等，另外，第二待推送内容可以包括具体内容，也可以仅包括主题内容，在此不做限定。在一些实施方式中，第一待推送内容和第二待推送内容的类型对应，也就是说，当第一待推送内容为广告时，那么第二待推送内容为广告；当第一待推送内容为游戏时，那么第二待推送内容为游戏。在一些实施方式中，第一待推送内容和第二待推送内容的类型不对应，例如第一待推送内容为游戏，第二待推送为包含游戏的文章等，在此不做限定。
例如，第一待推送内容为某汽车一子品牌的相应内容，通过第一待推送内容的词向量可以扩展出与该汽车一子品牌的相似度满足相似度阈值的其他子品牌的汽车品牌，将其他子品牌的汽车品牌确定为第二待推送内容。又例如，第一待推送内容为某游戏厂商推出的一游戏的相应内容，通过第一待推送内容的词向量可以扩展出与该游戏厂商推出的一游戏的相似度满足相似度阈值的其他游戏，将与该游戏的相似度满足相似度阈值的其他游戏确定为第二待推送内容。
步骤S130：确定与所述第一待推送内容对应的第一待推送用户，并确定与所述第二待推送内容对应的第二待推送用户。
在一些实施方式中，服务器在获取第一待推送内容后，可以确定与第一待推送内容对应的第一待推送用户，以及在获取第二待推送内容后，可以确定与第二待推送内容对应的第二待推送用户，其中，第一待推送用户和第二待推送用户中可以包括相同的用户，也可以包括不同的用户，在此不做限定。具体地，服务器可以先确定多个用户，并获取多个用户中的每个用户的属性信息，基于每个用户的属性信息，从多个用户中确定属性信息与第一待推送内容对应的用户作为第一待推送用户，以及从多个用户中确定属性信息与第二待推送内容对应的用户作为第二待推送用户。其中，属性信息可以包括兴趣标签、年龄信息、性别信息等，在此不做限定。
以属性信息为兴趣标签为例，当第一待推送内容为游戏A，第二待推送内容为游戏B时，服务器可以获取多个用户中的每个用户的兴趣标签，并判断每个用户的兴趣标签是否与游戏A对应，以及判断每个用户的兴趣标签是否与游戏B对应。其中，当用户的兴趣标签与游戏A对应时，可以将兴趣标签与游戏A对应的用户确定为第一待推送用户，当用户的兴趣标签与游戏B对应时，可以将兴趣标签与游戏B对应的用户确定为第二待推送用户。
步骤S140：将所述第一待推送用户和所述第二待推送用户共同确定为目标待推送用户。
在一些实施方式中，在确定与第一待推送内容对应的第一待推送用户和与第二待推送内容对应的第二待推送用户后，可以将第一待推送用户和第二待推送用户共同确定为目标待推送用户，将目标待推送用户确定为第一待推送内容的最终推送用户，以扩大第一待推送内容所推送的用户群体，提升第一待推送内容的推送效果。
本申请一个实施例提供的推送用户确定方法，获取第一待推送内容，并提取第一推送内容的词向量，获取与第一推送内容的词向量的相似度满足相似度阈值的第二待推送内容，确定与第一待推送内容对应的第一待推送用户，并确定与第二待推送内容对应的第二待推送用户，将第一待推送用户和第二待推送用户共同确定为目标待推送用户，从而通过提取第一待推送内容的词向量来获取相似度满足相似度阈值的第二待推送内容，并将第一待推送内容对应的第一待推送用户和第二待推送内容对应的第二待推送用户共同确定为目标待推送用户，以扩大内容推送的用户群体，提升内容推送效果。
请参阅图3，图3示出了本申请又一个实施例提供的推送用户确定方法的流程示意图。该方法应用于服务器，下面将针对图3所示的流程进行详细的阐述，所述推送用户确定方法具体可以包括以下步骤：
步骤S210：获取所述第一待推送内容，并确定所述第一待推送内容中的关键词。
在一些实施方式中，在获取第一待推送内容后，可以对第一待推送内容进行识别并提取该第一待推送内容中的关键词。在本实施例中，从第一待推送内容中提取的关键词数量可以为一个或多个，当从第一待推送内容中提取到的关键词的数量为一个时，可以直接将该关键词确定为第一待推送内容中的关键词；当从第一待推送内容中提取到的关键词的数量为多个时，可以将多个关键词均确定为第一待推送内容中的关键词，或者当从第一待推送内容中提取到的关键词的数量为多个时，可以对多个关键词进行筛选，从多个关键词中筛选比较重要的目标关键词，将目标关键词确定为第一待推送内容中的关键词。
步骤S220：将所述关键词转化为所述词向量。
在一些实施方式中，在确定第一待推送内容中的关键词后，可以将所确定的关键词转化为词向量，其 中，转化获得的词向量可以作为第一待推送内容的词向量。
步骤S230：分别计算所述第一待推送内容和多个待验证内容的词向量相似度。
在一些实施方式中，服务器在确定第一待推送内容的词向量之后，可以基于第一待推送内容的词向量分别计算第一待推送内容和多个待验证内容的词向量相似度。其中，在本实施例中，多个待验证内容可以为预先创建的资源池中的内容，另外，多个待验证内容可以包括广告、游戏、文章、音频、视频、链接等，在此不做限定。
作为一种方式，多个待验证内容可以为资源池中在预设时间段内比较受欢迎的内容，例如，可以为资源池中在预设时间段内的转化率高于指定转化率的内容，其中，转化率高于指定转化率可以包括：下载次数和浏览次数的比例高于第一指定比例，安装次数和浏览次数的比例高于第二指定比例，跳转次数和浏览次数的比例高于第三指定比例等，在此不做限定。
作为另一种方式，多个待验证内容可以为资源池中参考价值高于指定参考价值的内容，其中，参考价值高于指定参考价值可以包括：总用户量高于指定用户量、总评论数高于指定评论数、总得分高于指定总分、平均得分高于指定平均分等，在此不做限定。
请参阅图4，图4示出了本申请的图3所示的推送用户确定方法的步骤S230的一个实施例的流程示意图。下面将针对图4所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S231A：分别提取所述多个待验证内容中的每个待验证内容的词向量。
在一些实施方式中，在确定多个待验证内容后，可以分别提取多个待验证内容中的每个待验证内容的词向量。其中，可以通过词向量(word embedding)技术分别提取多个待验证内容中的每个待验证内容的词向量。于本实施例中，可以分别确定每个待验证内容中的关键词，并将每个待验证内容中的关键词转化为词向量。
步骤S232A：基于所述第一待推送内容的词向量和所述每个待验证内容的词向量，分别计算所述第一待推送内容和所述多个待验证内容的词向量相似度。
进一步地，在获取第一待推送内容的词向量和每个待验证内容的词向量之后，可以基于第一待推送内容的词向量和每个待验证内容的词向量，计算第一待推送内容和多个待验证内容中的每个待验证内容的词向量相似度。
请参阅图5，图5示出了本申请的图3所示的推送用户确定方法的步骤S230的又一个实施例的流程示意图。下面将针对图5所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S231B：将所述第一待推送内容的词向量输入预设的词向量模型，所述预设的词向量模型由所述多个待验证内容的词向量生成。
在一些实施方式中，服务器在获取第一待推送内容的词向量后，可以将该第一待推送内容的词向量输入词向量模型，其中，该词向量模型是由多个待验证内容的词向量生成。具体地，首先采集训练数据集，其中，训练数据集中的一类数据的属性或特征区别于另一类数据，然后通过将采集的训练数据集进行词向量相似度计算，从而基于该训练数据集总结出规律，得到词向量模型。在本实施例中，训练数据集的一类数据是多个待验证内容的词向量和参考推送内容的词向量，作为自变量，另一类数据是多个待验证内容的和参考推送内容的词向量相似度，作为因变量，对多个待验证内容的词向量和参考推送内容的词向量进行词向量相似度计算，可以得到多个待验证内容和参考推送内容之间的相关关系，进而获得词向量模型。
步骤S232B：获取所述预设的词向量模型输出的所述第一待推送内容和所述多个待验证内容的词向量相似度。
在一些实施方式中，词向量模型在接收到服务器输入的第一待推送内容的词向量后，可以对第一待推送内容的词向量进行处理，并输出第一待推送内容和多个待验证内容中的每个待验证推送的词向量相似度，相应的，服务器可以获取该词向量模型输出的第一待推送内容和多个待验证内容的词向量相似度。
请参阅图6，图6示出了本申请的图3所示的推送用户确定方法的步骤S230的再一个实施例的流程示意图。下面将针对图6所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S231C：获取所述第一待推送内容的主题内容。
在本实施例中，服务器在获取第一待推送内容后，可以对第一待推送内容进行分析，获取该第一待推送内容的主题内容。在一些实施方式中，服务器可以对第一待推送内容进行关键词提取，基于提取的关键词确定第一待推送内容的主题内容，例如，当提取到的关键词为“XX游戏”时，可以确定该第一待推送内容的主题内容为游戏；当提取到的关键词为“XX电影”时，可以确定该第一待推送内容的主题内容为电影。在一些实施方式中，服务器可以对第一待推送内容进行分词处理，并确定分词处理后的每个词组所出现的次数，获取出现次数最多的词组，并基于出现次数最多的词组确定第一待推送内容的主题内容，例如，当出现次数最多的词组为“XX游戏”时，可以确定该第一待推送内容的主题内容为游戏；当出现次数最多的词组为“XX电影”时，可以确定该第一待推送内容的主题内容为电影。
步骤S232C：获取与所述第一待推送内容的主题内容相同的多个待验证内容。
在一些实施方式中，在确定第一待推送内容的主题内容后，可以基于该第一待推送内容的主题内容从资源池包括的所有内容中，查找与该第一待推送内容的主题内容相同的内容，并将查找到的内容确定为待验证内容。例如，若第一待推送内容的主题内容为游戏，那么，可以从资源池包括的所有内容中查询主题内容为游戏相关的内容，并将查找到的主题内容为游戏相关的内容确定为待验证内容。
步骤S233C：分别计算所述第一待推送内容和所述多个待验证内容的词向量相似度。
请参阅图7，图7示出了本申请的图3所示的推送用户确定方法的步骤S230的另一个实施例的流程示意图。下面将针对图7所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S231D：获取所述第一待推送内容的类型。
在本实施例中，服务器在获取第一待推送内容后，可以对第一待推送内容进行分析，获取该第一待推送内容的类型。在一些实施方式中，服务器可以预先建立推送内容和类型之间的映射关系，并生成映射关系表存储在服务器中，如表1所示。其中，推送内容和类型之间的映射关系可以由用户手动关联，可以由服务器自动关联等，在此不做限定，并且推送内容和类型之间的映射关系可以包括一个推送内容对应一个类型，多个推送内容对应一个类型，一个推送内容对应多个类型等，在此不做限定。
进一步地，服务器在获取第一待推送内容后，可以将第一待推送内容和映射关系表中预先存储的多个推送内容一一对比，以获取与第一待推送内容匹配的推送内容，再根据所述映射关系表，查找该推送内容对应的类型，从而可以获取第一待推送内容的类型。例如，假设第一待推送内容为第一待推送应用，推送内容为推送应用，那么，当所述映射关系表中可以包括微信和即时通讯类的映射关系，也就是说，映射关系表中的推送应用下存储有微信，以及类型下存储有即时通讯类，微信和即时通讯类相关联，那么，当第一待推送应用为微信时，可以将第一待推送应用与映射关系表中存储的推送应用进行匹配，可以理解的，该第一待推送应用可以与映射关系表中的微信匹配，可以查找到该第一待推送应用的类型为即时通讯类。
表1
推送内容 类型
A1 B1
A2 B2
A3 B3
步骤S232D：获取与所述第一待推送内容的类型相同的多个待验证内容。
在一些实施方式中，在确定第一待推送内容的类型后，可以基于该第一待推送内容的类型从资源池包括的所有内容中，查找与该第一待推送内容的类型相同的内容，并将查找到的内容确定为待验证内容。例如，若第一待推送内容的类型为游戏类型，那么，可以从资源池包括的所有内容中查询类型为游戏类型的内容，并将查找到的类型为游戏类型的内容确定为待验证内容。
步骤S233D：分别计算所述第一待推送内容和所述多个待验证内容的词向量相似度。
步骤S240：从所述多个待验证内容中获取与所述第一待推送内容的词向量相似度大于指定相似度的目标待验证内容。
在一些实施方式中，服务器可以预先设置并存储指定相似度，该指定相似度用于作为每个待验证内容和第一待推送内容的词向量相似度的判断依据。其中，当待验证内容和第一待推送内容的词向量相似度大于指定相似度时，表征待验证内容和第一待推送内容的词向量相似度较高，即表征待验证内容和第一待推送内容较为类似，该待验证内容可以作为第二待推送内容。当待验证内容和第一待推送内容的词向量相似度不大于指定相似度时，表征待验证内容和第一待推送内容的词向量相似度较低，即表征待验证内容和第一待推送内容不类似，该待验证内容不可以作为第二待推送内容。
因此，在本实施例中，在获得第一待推送内容和每个待验证内容的词向量相似度后，可以分别将第一待推送内容和每个待验证内容的词向量与指定词向量相似度进行比较，以分别判断多个待验证内容和第一待推送内容的词向量相似度是否大于指定相似度，获得判断结果，并基于判断结果从多个待验证内容中获取与第一待推送内容的词向量相似度大于指定相似度的待验证内容，将与第一待推送内容的词向量相似度大于指定相似度的待验证内容确定为目标待验证内容。
步骤S250：将所述目标待验证内容确定为所述第二待推送内容。
步骤S260：确定与所述第一待推送内容对应的第一待推送用户，并确定与所述第二待推送内容对应的第二待推送用户。
其中，步骤S260的具体描述请参阅步骤S130，在此不再赘述。
请参阅图8，图8示出了本申请的图3所示的推送用户确定方法的步骤S260的流程示意图。下面将针对图8所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S261：获取多个用户对多个内容的当前兴趣标签，所述多个内容包括所述第一待推送内容。
在一些实施方式中，服务器可以预先获取多个用户的用户画像存储在本地。其中，该用户画像可以用于表征用户的兴趣标签、支付行为、推送转化率等，例如，用户画像可以用于表征目标用户偏好的游戏类型、偏好的视频类型、购买力、对推送内容的点击率、对推送内容的下载率等。
进一步地，服务器可以基于用户的用户画像确定用户针对包括第一待推送内容的多个内容的当前兴趣标签，其中，兴趣标签可以用于表征用户对多个内容的兴趣程度，其中，兴趣程度可以包括十分感兴趣、很感兴趣、比较感兴趣、或者不感兴趣等，在此不做限定。例如，可以获取用户针对第一待推送内容的兴趣程度，确定用户针对第一待推送内容是十分感兴趣、很感兴趣、比较感兴趣还是不感兴趣等。
请参阅图9，图9示出了本申请的图8所示的推送用户确定方法的步骤S261的流程示意图。下面将针对图9所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S2611：获取所述多个用户对所述多个内容的历史兴趣标签，并获取所述多个用户对所述多个内容的时间衰减因子。
在一些实施方式中，服务器可以分别记录多个用户对多个内容的历史行为，并基于多个用户对多个内容的历史行为确定多个用户对多个内容的历史兴趣标签。另外，服务器可以分别记录多个用户对多个内容的兴趣变化，并基于多个用户对多个内容的兴趣变化确定多个用户对多个内容的时间衰减因子，其中，用户对内容的兴趣变化可以包括用户对内容越来越感兴趣、越来越不感兴趣、在预设时段内感兴趣等，在此不做限定。
请参阅图10，图10示出了本申请的图9所示的推送用户确定方法的步骤S2611的一个实施例的流程示意图。下面将针对图10所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S26111A：获取所述多个用户在第一预设时间内对所述多个内容的浏览行为。
其中，用户对内容的浏览行为可以反映用户对内容的感兴趣程度。具体地，用户对内容的浏览行为越多，表征用户对内容越感兴趣，用户对内容的浏览行为越少，表征用户对内容越不感兴趣。于本实施例中，可以获取用户在第一预设时间内对内容的浏览行为，例如可以获取服务器记录的用户在第一预设时间内对内容的浏览次数。其中，第一预设时间为当前时刻之前的第一预设时长，所获取的用户在第一预设时间内对内容的浏览行为可以用于表征用户在当前时刻之前的第一预设时长对内容的感兴趣程度。在一些实施方式中，第一预设时间可以为1天、两天等，且第一预设时间可以由服务器自动设置，也可以由用户手动设置，在此不做限定。
步骤S26112A：基于所述浏览行为确定所述多个用户对所述多个内容的时间衰减因子。
其中，在获取用户在第一预设时间内对内容的浏览行为后，可以基于浏览行为确定用户对内容的时间衰减因子。例如，该浏览行为可以为浏览次数，则可以基于第一预设时间和浏览次数确定用户对内容的时间衰减因子。
请参阅图11，图11示出了本申请的图9所示的推送用户确定方法的步骤S2611的又一个实施例的流程示意图。下面将针对图11所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S26111B：获取所述多个用户在第二预设时间内对所述多个内容的下载行为。
其中，用户对内容的下载行为可以反映用户对内容的感兴趣程度。具体地，用户对内容的下载行为越多，表征用户对内容越感兴趣，用户对内容的下载行为越少，表征用户对内容越不感兴趣。于本实施例中，可以获取用户在第二预设时间内对内容的下载行为，例如可以获取服务器记录的用户在第二预设时间内对内容的下载次数，例如0次，1次等。其中，第二预设时间为当前时刻之前的第二预设时长，所获取的用户在第二预设时间内对内容的下载行为可以用于表征用户在当前时刻之前的第二预设时长对内容的感兴趣程度。在一些实施方式中，第二预设时间可以为1天、两天等，且第二预设时间可以由服务器自动设置，也可以由用户手动设置，在此不做限定。
步骤S26112B：基于所述下载行为确定所述多个用户对所述多个内容的时间衰减因子。
其中，在获取用户在第二预设时间内对内容的下载行为后，可以基于下载行为确定用户对内容的时间衰减因子。例如，该下载行为可以为下载次数，则可以基于第二预设时间和下载次数确定用户对内容的时间衰减因子。
请参阅图12，图12示出了本申请的图9所示的推送用户确定方法的步骤S2611的再一个实施例的流程示意图。下面将针对图12所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S26111C：获取所述多个用户在第三预设时间内对所述多个内容的跳转行为。
其中，用户对内容的跳转行为可以反映用户对内容的感兴趣程度。具体地，用户对内容的跳转行为越多，表征用户对内容越感兴趣，用户对内容的跳转行为越少，表征用户对内容越不感兴趣，其中，跳转行为可以表征用户对内容的阅读情况。于本实施例中，可以获取用户在第三预设时间内对内容的跳转行为，例如可以获取服务器记录的用户在第三预设时间内对内容的跳转次数。其中，第三预设时间为当前时刻之 前的第三预设时长，所获取的用户在第三预设时间内对内容的跳转行为可以用于表征用户在当前时刻之前的第三预设时长对内容的感兴趣程度。在一些实施方式中，第三预设时间可以为1天、两天等，且第三预设时间可以由服务器自动设置，也可以由用户手动设置，在此不做限定。
步骤S26112C：基于所述跳转行为确定所述多个用户对所述多个内容的时间衰减因子。
其中，在获取用户在第三预设时间内对内容的跳转行为后，可以基于跳转行为确定用户对内容的时间衰减因子。例如，该下载行为可以为跳转次数，则可以基于第三预设时间和跳转次数确定用户对内容的时间衰减因子。
步骤S2612：基于所述历史兴趣标签和时间衰减因子，获得所述多个用户对所述多个内容的当前兴趣标签。
在一些实施方式中，服务器在获取用户对内容的历史兴趣标签和时间衰减因子后，可以基于用户对内容的历史兴趣标签和时间衰减因子，确定用户对内容的当前兴趣标签。例如，若在之前用户对内容的历史兴趣标签表征用户对内容比较感兴趣，且时间衰减因子在下降，那么，用户对内容的当前兴趣标签可能从比较感兴趣变化为不感兴趣，即确定用户对内容的当前兴趣标签为不感兴趣。
在一些实施方式中，可以获取用户对内容的兴趣标签的排名。其中，用户对内容的兴趣标签的排名可以通过牛顿冷却定律获得，具体可以通过公式T＝T0e -λt来构建建立"热度"与"时间"之间如图13的指数衰减的过程，其中，λ即为时间衰减因子。基于时间衰减因子的用户兴趣标签，可以保留用户真正一直感兴趣的标签，而突发性的不能代表用户真正兴趣的热点标签最终会随时间逐渐衰减。
步骤S262：从所述多个用户中确定所述当前兴趣标签满足预设条件的目标用户作为所述第一待推送用户。
在一些实施方式中，服务器可以预先设置并存储预设条件，该预设条件用于作为用户对内容的当前兴趣标签的判断依据。其中，当兴趣标签满足预设条件时，表征用户对第一待推送内容的当前兴趣较高，即该用户可以作为第一待推送用户。当兴趣标签不满足预设条件时，表征用户对第一待推送内容的当前兴趣较低，即该用户不可以作为第一待推送用户。
因此，在本实施例中，在获得用户对多个内容的当前兴趣标签后，可以分别将当前兴趣标签和预设条件进行比较，以分别判断当前兴趣标签是否满足预设条件，当判断结果表征当前兴趣标签满足预设条件时，可以确定该当前兴趣标签对应的用户，并将该用户确定为第一待推送用户。
请参阅图14，图14示出了本申请的图8所示的推送用户确定方法的步骤S262的流程示意图。下面将针对图14所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S2621：获取所述目标用户针对所述第一待推送内容的当前热度值。
在一些实施方式中，基于目标用户针对第一待推送内容的当前兴趣标签，可以获取目标用户针对第一待推送内容的当前热度值。其中，热度值越高，表征目标用户对第一待推送内容越感兴趣，热度值越低，表征目标用户对第一待推送内容越不感兴趣。
步骤S2622：当所述当前热度值大于指定热度值时，将所述目标用户确定为所述第一待推送用户。
其中，服务器可以预先设置并存储指定热度值，该指定热度值可以用于作为目标用户对第一待推送内容的当前热度值的判断依据。因此，在本实施例中，当获取目标用户对第一待推送内容的当前热度值后，可以将当前热度值和指定热度值进行比较，以判断当前热度值是否大于指定热度值。其中，当判断结果表征当前热度值大于指定热度值时，表征目标用户当前对第一待推送内容的兴趣较高，可以将目标用户确定为第一待推送用户。
步骤S270：将所述第一待推送用户和所述第二待推送用户共同确定为目标待推送用户。
其中，步骤S270的具体描述请参阅步骤S140，在此不再赘述。
步骤S280：向所述目标待推送用户推送所述第一待推送内容。
进一步地，服务器在确定目标待推送用户后，可以将第一待推送内容推送给目标待推送用户，以扩大第一待推送内容所推送的用户群体。
请参阅图15，图15示出了本申请的图7所示的推送用户确定方法的步骤S280的一个实施例的流程示意图。下面将针对图15所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S281A：获取所述目标待推送用户的第一用户画像，所述第一用户画像用于表征所述目标待推送用户对于所展示内容的展示偏好。
在一些实施方式中，服务器可以预先获取目标待推送用户的第一用户画像存储在本地，在确定目标待推送用户后，可以从本地读取目标待推送用户的第一用户画像。于本实施例中，该第一用户画像用于表征目标待推送用户对于电子设备所展示内容的展示偏好，例如，用户画像可以用于表征目标待推送用户对于电子设备所展示内容的展示位置的偏好、所展示内容的展示大小的偏好、所展示内容的展示颜色的偏好、所展示内容的展示时长的偏好等，在此不做限定。
步骤S282A：基于所述目标待推送用户的第一用户画像，确定与所述第一用户画像对应的内容 展示方式。
在一些实施方式中，服务器在获取到目标待推送用户的第一用户画像后，可以根据目标待推送用户的第一用户画像确定与目标待推送用户的第一用户画像对应的内容展示方式。可以理解的是，根据目标待推送用户的第一用户画像所确定的内容展示方式与目标待推送用户对于电子设备所展示内容的展示偏好相对应。例如，若第一用户画像表征目标待推送用户对于电子设备所展示内容的展示位置的偏好为电子设备的左上方时，那么，根据目标待推送用户的第一用户画像确定的内容展示方式所对应的展示位置为电子设备的左上方。
在一些实施方式中，可以预先建立用户画像和内容展示方式的映射关系，该映射关系可以包括多个用户画像和多个内容展示方式，其中多个用户画像和多个内容展示方式可以一一对应、多个用户画像可以和一个内容展示方式对应、一个用户画像可以和多个内容展示方式对应等。在本实施例中，可以获取目标待推送用户的第一用户画像，再将第一用户画像和预先建立的映射关系中的多个用户画像进行对比，以获取与第一用户画像匹配的用户画像，再基于映射关系获取与第一用户画像匹配的用户画像对应的内容展示方式，将所获取的内容展示方式确定为第一用户画像对应的内容展示方式。
步骤S283A：基于所述内容展示方式确定所述第一待推送内容的推送格式，并按所述推送格式向所述目标待推送用户推送所述第一待推送内容。
在一些实施方式中，在确定与目标待推送用户的第一用户画像对应的内容展示方式后，可以基于所确定的内容展示方式确定第一待推送内容的推送格式，并按该推送格式向目标待推送用户推送该第一待推送内容，从而使得第一待推送内容最终在电子设备的展示方式更符合用户的喜好，可以提升推送内容的转化率和用户体验。
作为一种方式，在确定目标待推送用户的第一用户画像对应的内容展示方式后，可以基于该内容展示方式对推送内容进行调整，以使得调整后的推送内容的格式与内容展示方式一致。例如，若推送内容的字体大小为四号字体，所确定的内容展示方式的字体大小为五号字体，那么，可以将推送内容的字体大小从四号字体调整为五号字体，以使得调整后的推送内容的字体大小与内容展示方式的字体大小一致。
请参阅图16，图16示出了本申请的图7所示的推送用户确定方法的步骤S280的又一个实施例的流程示意图。下面将针对图16所示的流程进行详细的阐述，所述方法具体可以包括以下步骤：
步骤S281B：获取所述目标待推送用户的第二用户画像。
在一些实施方式中，服务器可以预先获取目标待推送用户的第二用户画像存储在本地，在确定目标待推送用户后，可以从本地读取目标待推送用户的第二用户画像。
步骤S282B：根据所述目标待推送用户的第二用户画像，确定对所述目标待推送用户进行限制的限制内容。
在本实施例中，服务器在获取到目标待推送用户的第二用户画像后，可以根据目标待推送用户的第二用户画像，确定对目标待推送用户进行限制的限制内容。在一些实施方式中，可以预先建立用户画像和限制内容的映射关系，如表2所示，该映射关系可以包括多个用户画像和多个限制内容，其中多个用户画像和多个限制内容可以一一对应、多个用户画像可以和一个限制内容对应、一个用户画像可以和多个限制内容对应等。在本实施例中，可以获取目标待推送用户的第二用户画像，再将目标待推送用户的第二用户画像和预先建立的映射关系中的多个用户画像进行对比，以获取与目标待推送用户的第二用户画像匹配的用户画像，再基于映射关系获取与目标待推送用户的第二用户画像匹配的用户画像对应的限制内容，在确定与目标待推送用户的第二用户画像匹配的用户画像对应的限制内容时，可以将该限制内容确定为对目标待推送用户进行限制的限制内容。
表2
用户画像 限制内容
用户画像1 限制内容1
用户画像2 限制内容2
用户画像3 限制内容3
在一些实施方式中，第二用户画像包括用户属性。基于此，在本实施例中，可以根据目标待推送用户的用户属性，确定对所述目标待推送用户进行限制的限制内容。其中，用户属性可以包括用户年龄、用户所在地、用户经历、用户性别等。其中，当用户属性包括用户年龄时，可以根据目标待推送用户的用户年龄，确定对所述目标待推送用户进行限制的限制内容。当用户属性包括用户所在地时，可以根据目标待推送用户的用户所在地，确定对所述目标待推送用户进行限制的限制内容。当用户属性包括用户经历时，可以根据目标待推送用户的用户经历，确定对所述目标待推送用户进 行限制的限制内容。当用户属性包括用户性别时，可以根据目标待推送用户的用户性别，确定对所述目标待推送用户进行限制的限制内容。
以用户属性包括用户所在地为例，用户位于不同的地方其所对应的需要进行限制的限制内容不同，例如，用户所在地为比较偏远的山区，则对目标待推送用户进行限制的限制内容可以包括外卖推送。以用户属性包括用户经历为例，用户经历不同其所对应的需要进行限制的限制内容不同，例如，用户经历不包括炒股经历，则对目标待推送用户进行限制的限制内容可以包括炒股推送。以用户属性包括用户年龄为例，用户年龄不同其所对应的需要进行限制的限制内容不同，例如，用户年龄过小(未成年)，则对目标待推送用户进行限制的限制内容可以包括游戏推送。
步骤S283B：当所述第一待推送内容中不包括所述限制内容时，向所述目标待推送用户推送所述第一待推送内容。
在一些实施方式中，在确定对目标待推送用户进行限制的限制内容后，可以判断第一待推送内容中是否包括限制内容，例如，可以将限制内容和第一待推送内容进行对比，以判断第一待推送内容中是否包括限制内容。
其中，当判断结果表征第一待推送内容中不包括限制内容时，可以向目标待推送用户推送第一待推送内容，以扩展第一待推送内容所推送的用户群体，
步骤S284B：当所述第一待推送内容中包括所述限制内容时，禁止向所述目标待推送用户推送所述第一待推送内容。
其中，当判断结果表征第一待推送内容中包括限制内容时，可以禁止向目标待推送用户推送第一待推送内容，以避免将不合适的内容推送给目标待推送用户，从而影响用户体验以及影响推送转化率的问题。
本申请又一个实施例提供的推送用户确定方法，获取第一待推送内容，并确定第一待推送内容中的关键词，将关键词转化为词向量，分别计算第一待推送内容和多个待验证内容的词向量相似度，从多个待验证内容中获取与第一待推送内容的词向量相似度大于指定相似度的目标待验证内容，将目标待验证内容确定为第二待推送内容。确定与第一待推送内容对应的第一待推送用户，并确定与第二待推送内容对应的第二待推送用户，将第一待推送用户和第二待推送用户共同确定为目标待推送用户，向目标待推送用户推送第一待推送内容。相较于图2所示的推送用户确定方法，本实施例还通过计算第一待推送内容和多个待验证内容的词向量相似度的方式确定与第一待推送内容的相似度满足相似度阈值的第二待推送内容，以提升相似内容获取的准确性。
请参阅图17，图17示出了本申请实施例提供的推送用户确定装置300的模块框图。该推送用户确定装置300应用于上述服务器，下面将针对图17所示的框图进行阐述，所述推送用户确定装置300包括：第一待推送内容获取模块310、第二待推送内容获取模块320、待推送用户确定模块330以及目标待推送用户确定模块340，其中：
第一待推送内容获取模块310，用于获取第一待推送内容，并提取所述第一待推送内容的词向量。
进一步地，所述第一待推送内容获取模块310包括：词向量相似度计算子模块、目标待验证内容获取子模块以及第二待推送内容确定子模块，其中：
词向量相似度计算子模块，用于分别计算所述第一待推送内容和多个待验证内容的词向量相似度。
进一步地，所述词向量相似度计算子模块包括：词向量提取单元和第一词向量相似度计算单元，其中：
词向量提取单元，用于分别提取所述多个待验证内容中的每个待验证内容的词向量。
第一词向量相似度计算单元，用于基于所述第一待推送内容的词向量和所述每个待验证内容的词向量，分别计算所述第一待推送内容和所述多个待验证内容的词向量相似度。
进一步地，所述词向量相似度计算子模块包括：词向量输入单元和第二词向量相似度计算单元，其中：
词向量输入单元，用于将所述第一待推送内容的词向量输入预设的词向量模型，所述预设的词向量模型由所述多个待验证内容的词向量生成。
第二词向量相似度计算单元，用于获取所述预设的词向量模型输出的所述第一待推送内容和所述多个待验证内容的词向量相似度。
进一步地，所述词向量相似度计算子模块包括：主题内容获取单元、第一待验证内容获取单元以及第三词向量相似度计算单元，其中：
主题内容获取单元，用于获取所述第一待推送内容的主题内容。
第一待验证内容获取单元，用于获取与所述第一待推送内容的主题内容相同的多个待验证内容。
第三词向量相似度计算单元，用于分别计算所述第一待推送内容和所述多个待验证内容的词向量相似度。
进一步地，所述词向量相似度计算子模块包括：类型获取单元、第二待验证内容获取单元以及第四词向量相似度计算单元，其中：
类型获取单元，用于获取所述第一待推送内容的类型。
第二待验证内容获取单元，用于获取与所述第一待推送内容的类型相同的多个待验证内容。
第四词向量相似度计算单元，用于分别计算所述第一待推送内容和所述多个待验证内容的词向量相似度。
目标待验证内容获取子模块，用于从所述多个待验证内容中获取与所述第一待推送内容的词向量相似度大于指定相似度的目标待验证内容。
第二待推送内容确定子模块，用于将所述目标待验证内容确定为所述第二待推送内容。
进一步地，所述第一待推送内容获取模块310包括：关键词确定子模块和词向量转化子模块，其中：
关键词确定子模块，用于获取所述第一待推送内容，并确定所述第一待推送内容中的关键词。
词向量转化子模块，用于将所述关键词转化为所述词向量。
第二待推送内容获取模块320，用于获取与所述第一待推送内容的词向量的相似度满足相似度阈值的第二待推送内容。
待推送用户确定模块330，用于确定与所述第一待推送内容对应的第一待推送用户，并确定与所述第二待推送内容对应的第二待推送用户。
进一步地，所述待推送用户确定模块330包括：当前兴趣标签获取子模块和第一待推送用户确定子模块，其中：
当前兴趣标签获取子模块，用于获取多个用户对多个内容的当前兴趣标签，所述多个内容包括所述第一待推送内容。
进一步地，所述当前兴趣标签获取子模块包括：时间衰减因子获取单元和当前兴趣标签获取单元，其中：
时间衰减因子获取单元，用于获取所述多个用户对所述多个内容的历史兴趣标签，并获取所述多个用户对所述多个内容的时间衰减因子。
进一步地，所述时间衰减因子获取单元包括：浏览行为获取子单元和第一时间衰减因子确定子单元，其中：
浏览行为获取子单元，用于获取所述多个用户在第一预设时间内对所述多个内容的浏览行为。
第一时间衰减因子确定子单元，用于基于所述浏览行为确定所述多个用户对所述多个内容的时间衰减因子。
进一步地，所述时间衰减因子获取单元包括：下载行为获取子单元和第二时间衰减因子确定子单元，其中：
下载行为获取子单元，用于获取所述多个用户在第二预设时间内对所述多个内容的下载行为。
第二时间衰减因子确定子单元，用于基于所述下载行为确定所述多个用户对所述多个内容的时间衰减因子。
进一步地，所述时间衰减因子获取单元包括：跳转行为获取子单元和第三时间衰减因子确定子单元，其中：
跳转行为获取子单元，用于获取所述多个用户在第三预设时间内对所述多个内容的跳转行为。
第三时间衰减因子确定子单元，用于基于所述跳转行为确定所述多个用户对所述多个内容的时间衰减因子。
当前兴趣标签获取单元，用于基于所述历史兴趣标签和时间衰减因子，获得所述多个用户对所述多个内容的当前兴趣标签。
第一待推送用户确定子模块，用于从所述多个用户中确定所述当前兴趣标签满足预设条件的目标用户作为所述第一待推送用户。
进一步地，所述第一待推送用户确定子模块包括：当前热度值获取单元和第一待推送用户确定单元，其中：
当前热度值获取单元，用于获取所述目标用户针对所述第一待推送内容的当前热度值。
第一待推送用户确定单元，用于当所述当前热度值大于指定热度值时，将所述目标用户确定为所述第一待推送用户。
目标待推送用户确定模块340，用于将所述第一待推送用户和所述第二待推送用户共同确定为目标待推送用户。
进一步地，所述推送用户确定装置300还包括：第一待推送内容推送模块，其中：
第一待推送内容推送模块，用于向所述目标待推送用户推送所述第一待推送内容。
进一步地，所述第一待推送内容推送模块包括：
第一用户画像获取子模块，用于获取所述目标待推送用户的第一用户画像，所述第一用户画像用于表征所述目标待推送用户对于所展示内容的展示偏好。
内容展示方式确定子模块，用于基于所述目标待推送用户的第一用户画像，确定与所述第一用户画像对应的内容展示方式。
推送格式确定子模块，用于基于所述内容展示方式确定所述第一待推送内容的推送格式，并按所述推送格式向所述目标待推送用户推送所述第一待推送内容。
进一步地，所述第一待推送内容推送模块包括：第二用户画像获取子模块、限制内容确定子模块、第一待推送内容推送子模块以及第一待推送内容禁止子模块，其中：
第二用户画像获取子模块，用于获取所述目标待推送用户的第二用户画像。
限制内容确定子模块，用于根据所述目标待推送用户的第二用户画像，确定对所述目标待推送用户进行限制的限制内容。
第一待推送内容推送子模块，用于当所述第一待推送内容中不包括所述限制内容时，向所述目标待推送用户推送所述第一待推送内容。
第一待推送内容禁止子模块，用于当所述第一待推送内容中包括所述限制内容时，禁止向所述目标待推送用户推送所述第一待推送内容。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述装置和模块的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
在本申请所提供的几个实施例中，模块相互之间的耦合可以是电性，机械或其它形式的耦合。
另外，在本申请各个实施例中的各功能模块可以集成在一个处理模块中，也可以是各个模块单独物理存在，也可以两个或两个以上模块集成在一个模块中。上述集成的模块既可以采用硬件的形式实现，也可以采用软件功能模块的形式实现。
请参阅图18，其示出了本申请实施例提供的一种服务器100的结构框图。本申请中的服务器100可以包括一个或多个如下部件：处理器110、存储器120以及一个或多个应用程序，其中一个或多个应用程序可以被存储在存储器120中并被配置为由一个或多个处理器110执行，一个或多个程序配置用于执行如前述方法实施例所描述的方法。
其中，处理器110可以包括一个或者多个处理核。处理器110利用各种接口和线路连接整个服务器100内的各个部分，通过运行或执行存储在存储器120内的指令、程序、代码集或指令集，以及调用存储在存储器120内的数据，执行服务器100的各种功能和处理数据。可选地，处理器110可以采用数字信号处理(Digital Signal Processing，DSP)、现场可编程门阵列(Field－Programmable Gate Array，FPGA)、可编程逻辑阵列(Programmable Logic Array，PLA)中的至少一种硬件形式来实现。处理器110可集成中央处理器(Central Processing Unit，CPU)、图形处理器(Graphics Processing Unit，GPU)和调制解调器等中的一种或几种的组合。其中，CPU主要处理操作***、用户界面和应用程序等；GPU用于负责待显示内容的渲染和绘制；调制解调器用于处理无线通信。可以理解的是，上述调制解调器也可以不集成到处理器110中，单独通过一块通信芯片进行实现。
存储器120可以包括随机存储器(Random Access Memory，RAM)，也可以包括只读存储器(Read-Only Memory)。存储器120可用于存储指令、程序、代码、代码集或指令集。存储器120可包括存储程序区和存储数据区，其中，存储程序区可存储用于实现操作***的指令、用于实现至少一个功能的指令(比如触控功能、声音播放功能、图像播放功能等)、用于实现下述各个方法实施例的指令等。存储数据区还可以存储终端100在使用中所创建的数据(比如电话本、音视频数据、聊天记录数据)等。
请参阅图19，其示出了本申请实施例提供的一种计算机可读存储介质的结构框图。该计算机可读介质400中存储有程序代码，所述程序代码可被处理器调用执行上述方法实施例中所描述的方法。
计算机可读存储介质400可以是诸如闪存、EEPROM(电可擦除可编程只读存储器)、EPROM、硬盘或者ROM之类的电子存储器。可选地，计算机可读存储介质400包括非易失性计算机可读介质(non-transitory computer-readable storage medium)。计算机可读存储介质400具有执行上述方法中的任何方法步骤的程序代码410的存储空间。这些程序代码可以从一个或者多个计算机程序产品中读出或者写入到这一个或者多个计算机程序产品中。程序代码410可以例如以适当形式进行压缩。
综上所述，本申请实施例提供的推送用户确定方法、装置、服务器以及存储介质，获取第一待推送内容，并提取第一推送内容的词向量，获取与第一推送内容的词向量的相似度满足相似度阈值的第二待推送内容，确定与第一待推送内容对应的第一待推送用户，并确定与第二待推送内容对应的第二待推送用户，将第一待推送用户和第二待推送用户共同确定为目标待推送用户，从而通过提取第一待推送内容的词向量来获取相似度满足相似度阈值的第二待推送内容，并将第一待推送内容对应的第一待推送用户和第二待推送内容对应的第二待推送用户共同确定为目标待推送用户，以扩大内容推送的用户群体，提升内容推送效果。
最后应说明的是：以上实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述实施例对本申请进行了详细的说明，本领域的普通技术人员当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不驱使相应技术方案的本质脱离本申请各实施例技术方案的精神和范围。

Claims (20)

  1. 一种推送用户确定方法，其特征在于，所述方法包括：
    获取第一待推送内容，并提取所述第一待推送内容的词向量；
    获取与所述第一待推送内容的词向量的相似度满足相似度阈值的第二待推送内容；
    确定与所述第一待推送内容对应的第一待推送用户，并确定与所述第二待推送内容对应的第二待推送用户；
    将所述第一待推送用户和所述第二待推送用户共同确定为目标待推送用户。
  2. 根据权利要求1所述的方法，其特征在于，所述获取与所述第一待推送内容的词向量的相似度满足相似度阈值的第二待推送内容，包括：
    分别计算所述第一待推送内容和多个待验证内容的词向量相似度；
    从所述多个待验证内容中获取与所述第一待推送内容的词向量相似度大于指定相似度的目标待验证内容；
    将所述目标待验证内容确定为所述第二待推送内容。
  3. 根据权利要求2所述的方法，其特征在于，所述分别计算所述第一待推送内容和多个待验证内容的词向量相似度，包括：
    分别提取所述多个待验证内容中的每个待验证内容的词向量；
    基于所述第一待推送内容的词向量和所述每个待验证内容的词向量，分别计算所述第一待推送内容和所述多个待验证内容的词向量相似度。
  4. 根据权利要求2所述的方法，其特征在于，所述分别计算所述第一待推送内容和多个待验证内容的词向量相似度，包括：
    将所述第一待推送内容的词向量输入预设的词向量模型，所述预设的词向量模型由所述多个待验证内容的词向量生成；
    获取所述预设的词向量模型输出的所述第一待推送内容和所述多个待验证内容的词向量相似度。
  5. 根据权利要求2-4任一项所述的方法，其特征在于，所述分别计算所述第一待推送内容和多个待验证内容的词向量相似度，包括：
    获取所述第一待推送内容的主题内容；
    获取与所述第一待推送内容的主题内容相同的多个待验证内容；
    分别计算所述第一待推送内容和所述多个待验证内容的词向量相似度。
  6. 根据权利要求2-4任一项所述的方法，其特征在于，所述分别计算所述第一待推送内容和多个待验证内容的词向量相似度，包括：
    获取所述第一待推送内容的类型；
    获取与所述第一待推送内容的类型相同的多个待验证内容；
    分别计算所述第一待推送内容和所述多个待验证内容的词向量相似度。
  7. 根据权利要求1-6任一项所述的方法，其特征在于，所述获取第一待推送内容，并提取所述第一待推送内容的词向量，包括：
    获取所述第一待推送内容，并确定所述第一待推送内容中的关键词；
    将所述关键词转化为所述词向量。
  8. 根据权利要求1-7任一项所述的方法，其特征在于，所述确定与所述第一待推送内容对应的第一待推送用户，包括：
    获取多个用户对多个内容的当前兴趣标签，所述多个内容包括所述第一待推送内容；
    从所述多个用户中确定所述当前兴趣标签满足预设条件的目标用户作为所述第一待推送用户。
  9. 根据权利要求8所述的方法，其特征在于，所述从所述多个用户中确定所述当前兴趣标签满足预设条件的目标用户作为所述第一待推送用户，包括：
    获取所述目标用户针对所述第一待推送内容的当前热度值；
    当所述当前热度值大于指定热度值时，将所述目标用户确定为所述第一待推送用户。
  10. 根据权利要求8或9所述的方法，其特征在于，所述获取多个用户对多个内容的当前兴趣标签，包括：
    获取所述多个用户对所述多个内容的历史兴趣标签，并获取所述多个用户对所述多个内容的时间衰减因子；
    基于所述历史兴趣标签和时间衰减因子，获得所述多个用户对所述多个内容的当前兴趣标签。
  11. 根据权利要求10所述的方法，其特征在于，所述获取所述多个用户对所述多个内容的时间衰减因子，包括：
    获取所述多个用户在第一预设时间内对所述多个内容的浏览行为；
    基于所述浏览行为确定所述多个用户对所述多个内容的时间衰减因子。
  12. 根据权利要求10或11所述的方法，其特征在于，所述获取所述多个用户对所述多个内容的时间衰减因子，包括：
    获取所述多个用户在第二预设时间内对所述多个内容的下载行为；
    基于所述下载行为确定所述多个用户对所述多个内容的时间衰减因子。
  13. 根据权利要求10-12任一项所述的方法，其特征在于，所述获取所述多个用户对所述多个内容的时间衰减因子，包括：
    获取所述多个用户在第三预设时间内对所述多个内容的跳转行为；
    基于所述跳转行为确定所述多个用户对所述多个内容的时间衰减因子。
  14. 根据权利要求1-13任一项所述的方法，其特征在于，所述将所述第一待推送用户和所述第二待推送用户共同确定为目标待推送用户之后，还包括：
    向所述目标待推送用户推送所述第一待推送内容。
  15. 根据权利要求14所述的方法，其特征在于，所述向所述目标待推送用户推送所述第一待推送内容，包括：
    获取所述目标待推送用户的第一用户画像，所述第一用户画像用于表征所述目标待推送用户对于所展示内容的展示偏好；
    基于所述目标待推送用户的第一用户画像，确定与所述第一用户画像对应的内容展示方式；
    基于所述内容展示方式确定所述第一待推送内容的推送格式，并按所述推送格式向所述目标待推送用户推送所述第一待推送内容。
  16. 根据权利要求15所述的方法，其特征在于，所述向所述目标待推送用户推送所述第一待推送内容，包括：
    获取所述目标待推送用户的第二用户画像；
    根据所述目标待推送用户的第二用户画像，确定对所述目标待推送用户进行限制的限制内容；
    当所述第一待推送内容中不包括所述限制内容时，向所述目标待推送用户推送所述第一待推送内容。
  17. 根据权利要求16所述的方法，其特征在于，所述方法还包括：
    当所述第一待推送内容中包括所述限制内容时，禁止向所述目标待推送用户推送所述第一待推送内容。
  18. 一种推送用户确定装置，其特征在于，所述装置包括：
    第一待推送内容获取模块，用于获取第一待推送内容，并提取所述第一待推送内容的词向量；
    第二待推送内容获取模块，用于获取与所述第一待推送内容的词向量的相似度满足相似度阈值的第二待推送内容；
    待推送用户确定模块，用于确定与所述第一待推送内容对应的第一待推送用户，并确定与所述第二待推送内容对应的第二待推送用户；
    目标待推送用户确定模块，用于将所述第一待推送用户和所述第二待推送用户共同确定为目标待推送用户。
  19. 一种服务器，其特征在于，包括存储器和处理器，所述存储器耦接到所述处理器，所述存储器存储指令，当所述指令由所述处理器执行时所述处理器执行如权利要求1-17任一项所述的方法。
  20. 一种计算机可读取存储介质，其特征在于，所述计算机可读取存储介质中存储有程序代码，所述程序代码可被处理器调用执行如权利要求1-17任一项所述的方法。
PCT/CN2019/118100 2019-11-13 2019-11-13 推送用户确定方法、装置、服务器以及存储介质 WO2021092803A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN201980099734.0A CN114303351A (zh) 2019-11-13 2019-11-13 推送用户确定方法、装置、服务器以及存储介质
PCT/CN2019/118100 WO2021092803A1 (zh) 2019-11-13 2019-11-13 推送用户确定方法、装置、服务器以及存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2019/118100 WO2021092803A1 (zh) 2019-11-13 2019-11-13 推送用户确定方法、装置、服务器以及存储介质

Publications (1)

Publication Number Publication Date
WO2021092803A1 true WO2021092803A1 (zh) 2021-05-20

Family

ID=75911435

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/118100 WO2021092803A1 (zh) 2019-11-13 2019-11-13 推送用户确定方法、装置、服务器以及存储介质

Country Status (2)

Country Link
CN (1) CN114303351A (zh)
WO (1) WO2021092803A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113489771A (zh) * 2021-06-30 2021-10-08 平安普惠企业管理有限公司 推送信息的方法、装置、终端及存储介质
CN114549078A (zh) * 2022-02-22 2022-05-27 中国平安人寿保险股份有限公司 基于时序的客户行为处理方法、装置及相关设备

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102609500A (zh) * 2012-02-01 2012-07-25 北京百度网讯科技有限公司 一种问题推送方法和采用该方法的问答***和搜索引擎
CN104915354A (zh) * 2014-03-12 2015-09-16 深圳市腾讯计算机***有限公司 多媒体文件推送方法及装置
CN107122423A (zh) * 2017-04-06 2017-09-01 深圳Tcl数字技术有限公司 影视推介方法及装置
CN109241415A (zh) * 2018-08-20 2019-01-18 平安科技（深圳）有限公司 项目推荐方法、装置、计算机设备及存储介质
US20190147058A1 (en) * 2017-11-16 2019-05-16 Baidu Online Network Technology (Beijing) Co., Ltd. Method and apparatus for pushing multimedia content
CN110008396A (zh) * 2018-11-28 2019-07-12 阿里巴巴集团控股有限公司 对象信息推送方法、装置、设备及计算机可读存储介质
CN110413875A (zh) * 2019-06-26 2019-11-05 腾讯科技（深圳）有限公司 一种文本信息推送的方法以及相关装置

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105279204B (zh) * 2014-07-25 2019-04-09 阿里巴巴集团控股有限公司 信息推送方法和装置
CN104615655B (zh) * 2014-12-31 2019-04-23 小米科技有限责任公司 信息推荐方法和装置
CN107888671A (zh) * 2017-10-30 2018-04-06 阿里巴巴集团控股有限公司 一种信息推送方法和装置

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102609500A (zh) * 2012-02-01 2012-07-25 北京百度网讯科技有限公司 一种问题推送方法和采用该方法的问答***和搜索引擎
CN104915354A (zh) * 2014-03-12 2015-09-16 深圳市腾讯计算机***有限公司 多媒体文件推送方法及装置
CN107122423A (zh) * 2017-04-06 2017-09-01 深圳Tcl数字技术有限公司 影视推介方法及装置
US20190147058A1 (en) * 2017-11-16 2019-05-16 Baidu Online Network Technology (Beijing) Co., Ltd. Method and apparatus for pushing multimedia content
CN109241415A (zh) * 2018-08-20 2019-01-18 平安科技（深圳）有限公司 项目推荐方法、装置、计算机设备及存储介质
CN110008396A (zh) * 2018-11-28 2019-07-12 阿里巴巴集团控股有限公司 对象信息推送方法、装置、设备及计算机可读存储介质
CN110413875A (zh) * 2019-06-26 2019-11-05 腾讯科技（深圳）有限公司 一种文本信息推送的方法以及相关装置

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113489771A (zh) * 2021-06-30 2021-10-08 平安普惠企业管理有限公司 推送信息的方法、装置、终端及存储介质
CN114549078A (zh) * 2022-02-22 2022-05-27 中国平安人寿保险股份有限公司 基于时序的客户行为处理方法、装置及相关设备
CN114549078B (zh) * 2022-02-22 2024-05-14 中国平安人寿保险股份有限公司 基于时序的客户行为处理方法、装置及相关设备

Also Published As

Publication number Publication date
CN114303351A (zh) 2022-04-08

Similar Documents

Publication Publication Date Title
US11290775B2 (en) Computerized system and method for automatically detecting and rendering highlights from streaming videos
WO2018072513A1 (zh) 一种广告信息推送方法、装置以及***、服务器、计算机可读介质
CN110325986B (zh) 文章处理方法、装置、服务器及存储介质
US10332184B2 (en) Personalized application recommendations
KR20180114236A (ko) 콘텐츠 성능 최적화를 위한 자동화된 클릭 타입 선택
US10783431B2 (en) Image search using emotions
CN107423308B (zh) 主题推荐方法以及装置
US20210326400A1 (en) Method and apparatus for processing web content, device, and computer storage medium
TWI662495B (zh) 推廣資訊的處理方法、裝置及系統
WO2020108125A1 (zh) 一种信息推送方法、信息推送装置及终端设备
CN115212561B (zh) 基于玩家的语音游戏数据的服务处理方法及相关产品
US11948558B2 (en) Messaging system with trend analysis of content
JP6767342B2 (ja) 検索装置、検索方法および検索プログラム
WO2021092803A1 (zh) 推送用户确定方法、装置、服务器以及存储介质
JP2019036191A (ja) 判定装置、判定方法及び判定プログラム
WO2021142719A1 (zh) 画像生成方法、装置、服务器及存储介质
WO2018149244A1 (zh) 一种基于目标应用推荐相关联应用的方法和装置
CN109299378B (zh) 搜索结果的展示方法、装置、终端及存储介质
WO2021087746A1 (zh) 推送内容的处理方法、装置、电子设备以及存储介质
CN114387041B (zh) 一种多媒体数据采集方法及***
WO2021087747A1 (zh) 推送内容的处理方法、装置、电子设备以及存储介质
WO2017157130A1 (zh) 推送增值业务信息的方法、装置及电子设备
US11776011B2 (en) Methods and apparatus for improving the selection of advertising
US20210224310A1 (en) Electronic device and story generation method thereof
WO2017157159A1 (zh) 一种推送增值业务信息的方法、装置及电子设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19952534

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19952534

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 19952534

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205 DATED 28/10/2022)