WO2019095417A1

WO2019095417A1 - Real-time advertisement recommendation method and apparatus, and terminal device and storage medium

Info

Publication number: WO2019095417A1
Application number: PCT/CN2017/112569
Authority: WO
Inventors: 黄度新; 张川; 李双灵; 王翼; 金鑫
Original assignee: 平安科技（深圳）有限公司
Priority date: 2017-11-15
Filing date: 2017-11-23
Publication date: 2019-05-23
Also published as: CN107862553B; CN107862553A

Abstract

Disclosed are a real-time advertisement recommendation method and apparatus, and a terminal device and a storage medium. The real-time advertisement recommendation method comprises: acquiring an access request sent by a client in real time, wherein the access request comprises a request source identifier; based on the access request, acquiring user preference information corresponding to the request source identifier; based on the user preference information, acquiring an associated advertisement corresponding to the user preference information; and pushing the associated advertisement to the client in real time, so that the client displays the associated advertisement in real time. By means of the real-time advertisement recommendation method, an advertisement of interest to a user is recommended in real time according to user preference information, thereby improving the click-through rate of a pushed advertisement and achieving the purpose of advertisement pushing.

Description

Advertising real-time recommendation method, device, terminal device and storage medium

This patent application is based on the Chinese invention patent application filed on November 15, 2017, with the application number of 201711126787.6, entitled "Advertising real-time recommendation method, device, terminal device and storage medium", and requires its priority.

Technical field

The present application relates to the field of big data, and in particular, to a method, device, terminal device and storage medium for real-time advertisement recommendation.

Background technique

When the current user visits the website, the website will randomly push the advertisement to the user. Because it is randomly pushed, it is impossible to recommend the advertisement of the user's interest to the user in real time. When the user visits the website, he usually only browses the interest or himself and himself. The webpage or advertisement related to the demand, if the advertisement pushed by the website is an advertisement of interest to the user, the click rate of the user clicking the advertisement is high; on the contrary, if the advertisement pushed by the website is not the advertisement of the user, the user may not Clicking on the ad resulted in a lower click-through rate for push ads, and it did not achieve good ad push performance, and did not achieve the purpose of ad push.

Summary of the invention

The embodiment of the present application provides a method, a device, a terminal device and a storage medium for real-time advertisement recommendation, so as to solve the problem that the click rate of the current website random push advertisement is low.

In a first aspect, an embodiment of the present application provides a real-time advertisement recommendation method, including:

Acquiring an access request sent by the client in real time, where the access request includes a request source identifier;

Acquiring user preference information corresponding to the request source identifier based on the access request;

Acquiring an associated advertisement corresponding to the user preference information based on the user preference information;

Pushing the associated advertisement to the client in real time, so that the client displays the associated advertisement in real time.

In a second aspect, the embodiment of the present application provides an advertisement real-time recommendation device, including:

An access request obtaining module, configured to acquire an access request sent by the client in real time, where the access request includes a request source identifier;

a user preference information obtaining module, configured to acquire user preference information corresponding to the request source identifier based on the access request;

An associated advertisement obtaining module, configured to acquire an associated advertisement corresponding to the user preference information based on the user preference information;

The associated advertisement recommendation module is configured to push the associated advertisement to the client in real time, so that the client displays the associated advertisement in real time.

In a third aspect, an embodiment of the present application provides a terminal device, including a memory, a processor, and computer readable instructions stored in the memory and executable on the processor, where the processor executes the computer The steps of implementing the real-time recommendation method of the advertisement when reading the instruction:

In a fourth aspect, the embodiment of the present application provides a computer readable storage medium, where the computer readable storage medium stores computer readable instructions, and when the computer readable instructions are executed by a processor, the real-time recommendation method of the advertisement is implemented. step:

In the advertisement real-time recommendation method, device, terminal device and storage medium provided by the embodiment of the present application, the corresponding user preference information is obtained by acquiring the request source identifier in the access request sent by the client in real time, and then the corresponding association is obtained based on the user preference information. The advertisement, the associated advertisement is recommended to the corresponding client in real time, so that the client displays the associated advertisement that the user is interested in. The advertisement real-time recommendation method, the device, the terminal device and the storage medium can realize the related advertisements that are interested in the user in real time according to the user preference information, and improve the click rate of the push advertisement.

DRAWINGS

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings used in the description of the embodiments of the present application will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without the creative labor, it can also be obtained according to these drawings. Have other drawings.

1 is a flow chart of a method for real-time recommendation of advertisements in Embodiment 1 of the present application.

FIG. 2 is a specific schematic diagram of step S20 of FIG. 1.

FIG. 3 is another specific schematic diagram of step S20 of FIG. 1.

FIG. 4 is a specific schematic diagram of step S30 of FIG. 1.

FIG. 5 is a schematic block diagram of an advertisement real-time recommendation device in Embodiment 2 of the present application.

FIG. 6 is a schematic diagram of a terminal device in Embodiment 4 of the present application.

Detailed ways

The technical solutions in the embodiments of the present application are clearly and completely described in the following with reference to the drawings in the embodiments of the present application. It is obvious that the described embodiments are a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present application without departing from the inventive scope are the scope of the present application.

Example 1

FIG. 1 is a flow chart showing a method for recommending an advertisement in real time in this embodiment. The real-time recommendation method of the advertisement is applied when the user visits the website, and can recommend the advertisement of interest to the user in real time, thereby improving the click rate of the advertisement. As shown in FIG. 1, the real-time recommendation method of the advertisement includes the following steps:

S10: Obtain an access request sent by the client in real time, and the access request includes a request source identifier.

Specifically, the server connected to the client receives the access request sent by the client in real time, and the access request generally carries a URL address, and the server parses the URL address, and feeds back the content corresponding to the URL address to the client, so that the client Show the page. In this embodiment, the access request received by the server further includes a request source identifier, where the request source identifier is an identifier for uniquely identifying the source of the request. The request source identifier includes a user identifier and/or a terminal identifier; wherein the user identifier is an identifier for uniquely identifying a user that triggers the access request; the terminal identifier is an identifier for uniquely identifying a terminal that sends the access request. The user identifier may be account information input when the user logs in to a specific webpage, and the account information includes, but is not limited to, a login name, a mobile phone number, a micro signal, and an email address used when the user logs in. The terminal identifier includes, but is not limited to, an identifier such as a MAC address or an IP address of a terminal for uniquely identifying a source of an access request, such as a computer, a mobile phone, and an Ipad.

S20: Acquire, according to the access request, user preference information corresponding to the request source identifier.

Specifically, the server acquires the corresponding user preference information according to the request source identifier in the access request, and obtains the user image data corresponding to the request source identifier. User image data refers to the number of locations to a person According to data, identity data, consumption data, behavior data and lifestyle data, a tagged user model data abstracted by data analysis.

When the request source identifier of the access request is the user identifier, that is, when the user uses the account information to log in to the specific webpage, the access request sent by the user carries the user identifier (the terminal identifier can also be carried at the same time), and the server can be based on the The user identification looks up the user's portrait data of the user to obtain corresponding user preference information based on the found user image data.

When the user does not use the account information to log in to the specific webpage, that is, the user browses the specific webpage as the visitor, the access request received by the server does not carry the user identifier but only carries the terminal identifier, and the server can query the terminal identifier according to the terminal identifier. The terminal frequently accesses historical webpage data formed by the webpage to obtain corresponding user portrait data based on the historical webpage data, thereby determining corresponding user preference information. For example, according to the user browsing the red wine related webpage data through the terminal corresponding to the same terminal identifier for a period of time, by analyzing the behavior data of the user browsing the webpage, a label of the user may be abstracted, and the user likes red wine or recently needs to purchase. Red wine, and the favorite red wine is part of the user portrait data of the terminal corresponding to the terminal identification.

In summary, the user preference information may be information such as the personal preference of the user corresponding to the account information when the user logs in to the webpage by using the account information; or the user does not use the account information but visits the webpage as a visitor, according to the user Information such as personal preferences determined by historical web page data frequently accessed by the terminal. For example, a webpage frequently browsed by a user using the same terminal is a webpage related to red wine, and thus the user preference information may be inferred to be red wine; and, for example, a webpage frequently browsed by the same terminal is a webpage related to outdoor activity tourism, thereby It can be inferred that the user preference information is for outdoor travel.

In this embodiment, the user portrait data is acquired by the user portrait data system. The user portrait data system is divided into four subsystems: a data source subsystem, a data rotor system, a big data platform subsystem, and a data application subsystem.

The data source subsystem is mainly an application layer module, which is associated with the user and is used for data collection. Specifically, the data source subsystem can be divided into a data class module, an internet channel class module, and a third party data module. The data class module includes, but is not limited to, the core transaction module, the risk association module and the data warehouse module in the embodiment; the internet channel class module includes but is not limited to the portal website, the mobile banking and the WeChat bank in the embodiment; the third party data module This includes, but is not limited to, data from the extranet in this embodiment. Data is generated when a user logs in to a particular web page or browses a web page, and the data source subsystem is used to collect the data. In this embodiment, the data source subsystem mainly uses the distributed log real-time collection platform Flume for data collection, and sends the collected data to the distributed message middleware Kafka for aggregation. Finally, the distributed computing engine Spark is used for distributed messages. Middleware Kafka gets the data and processes the data.

The data in the rotor system is used to connect the data source subsystem and the big data platform subsystem, that is, to send the data acquired by the data source subsystem to the big data platform subsystem. In this embodiment, the data in the rotor system is used to collect data such as database files, transaction messages, system logs, and database logs collected by the data source subsystem, and send the data to the big data platform subsystem. For example, when a user logs in to a specific webpage or browses a webpage, the core transaction module converts the transaction record completed by the acquired user into a specific webpage, and triggers the relevant risk situation to generate a database file, a transaction message, a system log, and Data such as database logs, and the above data is sent to the big data platform subsystem. It can be understood that the rotor system in the data also functions as a data storage function for storing data uploaded by the data source subsystem. The distributed storage platform HBase is used for data storage. The storage platform HBase can realize the processes of network communication, message authentication, transaction data format conversion, personal password PIN conversion, transaction flow record, transaction preprocessing, transaction monitoring and transaction data statistics. The formed data is stored.

The Big Data Platform subsystem is used to process and calculate massive amounts of data. In this embodiment, the most important function of the big data platform subsystem is data calculation, and the data is analyzed by using Spark/Hive. For example, the user often browses the red wine related webpage information in a period of time browsing information, and the big data platform subsystem will Spark/Hive is used for data analysis of all online browsing information of the user, thereby obtaining user portrait data corresponding to the user identification or the terminal identification.

The data application subsystem is used to provide an interface for analysis results obtained by data analysis of the big data platform subsystem for other system calls, such as inputting the analysis result into an application system for data mining, deep learning, and data market. Specifically, the data application subsystem may perform data mining on information of users who like wine or purchase red wine to obtain data information such as age, gender, and geographic location of the user, and perform deep learning through the data information obtained by mining. To obtain user preference information similar to the user's age, gender, and geographic location. The data application subsystem can also obtain the age distribution of users who like red wine through data mining, where the geographical location is mainly concentrated, and according to the obtained data information combined with the data market application system, the key sales of places with high demand for red wine are Ad recommendation.

In a specific implementation, as shown in FIG. 2, in step S20, the user preference information corresponding to the request source identifier is obtained, which specifically includes the following steps:

S211: Determine, according to the access request, whether the request source identifier is a user identifier.

Specifically, the request source identifier carried in the access request received by the server may be a user identifier for uniquely identifying the user, or may be a terminal identifier for uniquely identifying the terminal, or carry the user identifier and the terminal identifier at the same time. The user identifier is an identifier carried by the access request formed by the user after logging in to the specific webpage by using the account information, and corresponds to the close-up user; and the terminal identifier is the identifier of the terminal that sends the access request, and is not limited to a specific user; therefore, the user identifier and the identifier are sent. The user of the access request is more closely contacted, so that when the server receives the access request in step S211, It is necessary to first determine whether the request source identifier in the access request is a user identifier for the user to access a specific webpage login.

S212: If the request source identifier is a user identifier, query the existing user portrait data based on the user identifier.

Among them, the existing user portrait data is user portrait data previously collected and stored in the database connected to the server and associated with the user identification. In this embodiment, the existing user portrait data includes, but is not limited to, basic information such as gender, age, region, address, occupation, marital status, consumption habits, and education level, and may also include preference information for embodying user preferences.

Specifically, when the user logs in to a specific webpage by using the account information, the server corresponding to the specific webpage forms a user access log, and the user access log may include gender, age, region, address, occupation, marital status, consumption habits, and personal preferences. Basic information such as the educational level, and may also include access information such as the user's access page, date of access, specific access time, and length of visit. The distributed log real-time collection platform Flume will collect user access logs from different servers in real time, and send the collected user access logs to the distributed message middleware Kafka for aggregation, so that each user access log is associated with the user ID. The distributed computing engine Spark obtains the user access log carrying the same user identifier from the distributed message middleware Kafka, and performs data processing on all the obtained user access logs, and labels the user to form user portrait data. Finally, the tagged user portrait data is stored in the distributed storage platform Hbase, and the user portrait data is stored in association with the user identifier, so that the corresponding existing user portrait data can be queried based on the user identifier. The above steps all adopt a distributed framework, which is beneficial for processing massive data and improving data processing efficiency.

For example, if the age, occupation, and personal preference information is carried in the user access log, the age obtained from the user access log may be used as a label of the user, and the acquired occupation is taken as another label of the user, and the obtained personal preference is obtained. As a user's further label... until all the basic information and/or access information in the collected user access log is tagged to obtain user portrait data, and the tagged user portrait data is stored in distributed storage. Platform Hbase. Any one of the user access logs may carry one or more basic information and/or access information corresponding to the user identifier, so that the acquired user portrait data carries at least one label, and the obtained user label is relatively wide, and the obtained label is used for the user. In fact, it is more accurate and targeted.

S213: Determine whether the existing user image data contains existing preference information.

Specifically, the user profile data includes all user tags associated with the user identification. In this embodiment, when the user logs in to a specific webpage by using the account information, the user may fill in one of the basic information including but not limited to gender, age, region, occupation, marital status, personal preference, and education level in the corresponding account information. Or multiple, in step S213, it is determined whether the existing user image data contains existing preference information.

S214: If the existing user image data includes the existing preference information, the existing preference information is used as the user preference information.

In this embodiment, if the existing user image data clearly determines that the existing preference information is included, the existing preference information in the existing user image data is directly used as the user preference information, so that the recommendation is subsequently performed based on the user preference information. Since the existing favorite information is mostly uploaded by the user, it is more suitable for the user's actual preference, and the existing favorite information is used as the basis for the advertisement recommendation, so that the pushed advertisement is more in line with the user's preference, so as to improve the advertisement to a certain extent. Clickthrough rate.

S215: If the existing user image data does not include the existing preference information, the similar person is searched based on the existing user portrait data.

Among them, the similar crowd is the most similar group of user portrait data and existing user portrait data. It can be understood that since the existing user image data does not include the existing preference information, it is necessary to find a similar crowd from the user portrait data system based on the existing user portrait data, so as to determine the corresponding user identifier based on the common preference information of the similar crowd. User preferences information for users.

In this embodiment, the big data platform subsystem uses Spark/Hive for data analysis, and clusters user image data of all users stored in the distributed storage platform Hbase to cluster all users according to their common preference information. . Specifically, user image data of all users may include, but is not limited to, gender, age, region, address, occupation, marital status, consumption habits, preference information, and education level. The K-means clustering algorithm is used to cluster the user portrait data of all users, so that all users are divided into several clustering groups based on the common preference information, and each clustering group corresponds to the clustering user portrait data. Among them, K-means clustering algorithm is a clustering algorithm based on distance evaluation similarity, that is, the closer the distance between two objects, the larger the similarity is. When determining the similar crowd based on the existing user portrait data, the Euclidean distance of the existing user portrait data and the cluster user image data corresponding to each cluster population is calculated, and the cluster population with the smallest Euclidean distance is selected as the similar crowd. Where the Euclidean distance of any two n-dimensional vectors a (xi1, xi2, ..., xin) and b (xj1, xj2, ..., xjn)

S216: The common preference information corresponding to the similar group is used as the user preference information.

Since the user portrait data of the similar group is the most similar to the existing user portrait data of the user corresponding to the user identifier, the preference is also most likely to be the same, so the common preference information of the similar group can be used as the user of the user corresponding to the user identifier. Like the information, and as the basis for the recommendation of the advertisement, the pushed advertisement is more in line with the user's preference, and the click rate of the advertisement is improved to some extent.

S217: If the existing user image data does not include the existing favorite information, the corresponding at least one historical webpage data is searched based on the user identifier, and each historical webpage data corresponds to a favorite label.

If the request source identifier in the access request is the user identifier but the existing user image data corresponding to the user identifier is not When the existing preference information is included, the server may search for at least one historical webpage data that the corresponding terminal has accessed according to the user identifier. The historical webpage data may be historical webpage data uploaded to the distributed storage platform Hbase. Since the historical webpage data is associated with the user identifier, it can be understood as a trace left by the user corresponding to the user identifier when accessing the webpage, so that the favorite tab corresponding to each historical webpage data can reflect the user's preference to a certain extent.

In this embodiment, each historical webpage data corresponds to a favorite tag, and the favorite tag can be obtained by using a Jieba word segmentation tool and a TF-IDF algorithm. Specifically, the Jieba word segmentation tool (ie, the word segmentation tool) scans the text information in the historical webpage data, and then divides the long words in the text information, and then performs part-of-speech tagging on the segmented text information to obtain Word segmentation results. Then, the TF-IDF algorithm is used to extract the keyword result of the segmentation result processed by the Jieba word segmentation tool, so that the extracted keyword is used as the favorite tag corresponding to the historical webpage data. Among them, using the TF-IDF algorithm to extract the keyword results of the word segmentation processed by the Jieba word segmentation tool includes the following steps:

First, calculate the word frequency (TF) of each word in the word segmentation result of any historical web page data. Where word frequency (TF) refers to the frequency at which a given word appears in the file, and its formula is

The numerator indicates the number of occurrences of the word in the file, and the denominator indicates the sum of the occurrences of all words in the file.

Then, the inverse document frequency (IDF) of each word in the word segmentation result of any historical web page data is calculated. Among them, the inverse document frequency (IDF) means that each word is assigned an "importance" weight, which means that the most common words (",""yes","at") give the smallest Weights, the more common words give less weight, the less common words give greater weight, this weight is called "inverse document frequency", and its size is inversely proportional to the common degree of a word. The inverse document frequency (IDF) formula can be expressed as:

Where |D| refers to the total number of files in the corpus, and |{j:t _u ∈d _j }| refers to the total number of files containing words.

Finally, TF-IDF _i,j =TF _i,j ×IDF _{i,j is used to} obtain the weight of each word in the historical webpage data, and the words with the highest weight or relatively high (ie, the first N digits) are selected as keywords. That is, the favorite tag corresponding to the history webpage data. TF-IDF tends to filter out common words, retain important words, and use the important words as keywords of the historical web page data. Select one keyword with the highest weight or several keywords with higher weight to determine the history page. The corresponding favorite tag in the data.

S218: Perform statistical analysis on the favorite tags corresponding to the historical webpage data, and obtain key preference tags to determine user preference information.

In step S217, the TF-IDF algorithm is used for each historical webpage data to perform keyword extraction to determine that each historical webpage data has a corresponding favorite label; in step S218, corresponding to all historical webpage data corresponding to the user identifier is required. The favorite tag is counted to determine the highest or higher favorite tag as the key preference tag, and the key preference tag is used as the finalized user preference information, so that the advertisement recommendation is based on the user preference information, so that the recommended advertisement is more Meet the interests of users and increase the click-through rate of your ads.

In a specific implementation, as shown in FIG. 3, in step S20, the user preference information corresponding to the request source identifier is obtained, which specifically includes the following steps:

S221: Determine, according to the access request, whether the request source identifier is a terminal identifier.

Specifically, the request source identifier carried in the access request received by the server may be a user identifier for uniquely identifying the user, or may be a terminal identifier for uniquely identifying the terminal, or carrying both the user identifier and the terminal identifier. When the user does not use the account information to log in to the specific webpage, the request source identifier in the access request received by the server is the terminal identifier, and the terminal identifier can uniquely determine the terminal that sends the access request.

S222: If the source identifier is the terminal identifier, the at least one historical webpage data is searched according to the terminal identifier, and each historical webpage data has a corresponding favorite label.

If the source identifier of the request in the access request is the terminal identifier, the server searches for at least one historical webpage data that the corresponding terminal has accessed according to the terminal identifier. The historical webpage data may be historical webpage data uploaded to the distributed storage platform Hbase, or may be history webpage data stored in a cookie (or a cookie) on the terminal. Among them, cookies (or cookies) refer to the data that some websites store on the user's local terminal in order to identify the user's identity and perform session tracking. The process of obtaining the user label in step S222 is similar to the process in step S217. To avoid repetition, details are not described herein.

S223: Perform statistical analysis on the favorite tags corresponding to the historical webpage data, and obtain key preference tags to determine user preference information.

In step S222, the TF-IDF algorithm is used for each historical webpage data to perform keyword extraction to determine that each historical webpage data has a corresponding favorite label; in step S223, all historical webpage data corresponding to the terminal identifier is required. The favorite tag is counted to determine the favorite tag with the highest frequency or high frequency as the key preference tag, and the key preference tag is used as the finalized user preference information, so that the recommendation is based on the user preference information, so that the recommended The advertisement is more in line with the user's interest and improves the click rate of the advertisement.

In a specific implementation, before performing the advertisement real-time recommendation method, in particular before step S20, the method further includes: labeling all the webpages on the website, so that each webpage carries a favorite label. It can be understood that the favorite tag of the webpage can be manually set by the webpage developer, or can be pre-adopted by the Jieba word segmentation tool and TF-IDF. The method processes the content of the webpage, and obtains keywords of the webpage content to determine corresponding favorite tags.

In this embodiment, step S20 specifically searches for a corresponding target webpage based on the URL address of the access request, and uses the favorite tab corresponding to the target webpage as the user preference information. The target webpage is a webpage corresponding to the URL address of the access request. Since all the webpages carry the favorite tags, the target webpage should also carry a corresponding favorite tag, and the favorite tag is used as the user preference information of the user who triggered the access request, so as to perform advertisement recommendation based on the user preference information, so that the recommendation is recommended. The ads are more in line with the user’s interest and increase the click-through rate of the ad. The determination of the user preference information based on the URL address is associated with the access request triggered by the user each time, and has great contingency. In the case that the user identifier is uncertain and the historical webpage data acquired based on the terminal identifier is small, the determined User preference information can largely reflect the user's true preferences. Recommending advertisements based on the user's favorite information can also effectively increase the click rate of the advertisement to a certain extent.

S30: Acquire an associated advertisement corresponding to the user preference information based on the user preference information.

The related advertisement refers to an advertisement whose content corresponds to the user preference information. After determining the user preference information corresponding to the request source identifier in the access request by using the step S20, the corresponding related advertisement may be searched based on the user preference information, so that the associated advertisement is more in line with the interest of the user who triggered the access request, so as to improve The user’s clickthrough rate for the associated ad.

In a specific implementation, as shown in FIG. 4, step S30 includes the following steps:

S31: Perform keyword extraction on the advertisement to determine an advertisement category of the advertisement.

The advertisement category refers to determining the category to which the advertisement belongs according to the advertisement content. Specifically, the advertising category includes, but is not limited to, travel advertisements, shopping advertisements, etc., and the travel advertisements may be subdivided into travel agency advertisements, hotel advertisements, tourist city/scenic area advertisements, travel festival celebration advertisements, and exhibition advertisements. In this embodiment, when the server obtains the advertisement to be pushed, the advertisement category may be determined based on the positioning of the advertisement by the advertiser, that is, the advertiser clearly determines the advertisement category; and the advertisement category may also be determined based on the advertisement content. When determining the advertisement category based on the advertisement content, the Jieba word segmentation tool and the TF-IDF algorithm may be used to process the advertisement content, and the keywords corresponding to the advertisement content are obtained to determine the advertisement category.

S32: Calculate the similarity between the advertisement category and the user preference information.

In this embodiment, the similarity between the advertisement category and the user preference information may be expressed by cosine similarity. The cosine similarity algorithm is used to calculate the advertisement category and user preference information. The calculation formula of the cosine similarity algorithm is

Where x is the weight corresponding to the keyword in the advertisement category, and y is the weight corresponding to each preference information in the user preference information. When the calculated cosine value is closer to 1, the advertisement category is proved to be similar to the user. The higher the information similarity, the more the advertising category is determined to be closer to the user preference information.

In a specific embodiment, advertisements can be classified according to industry categories, such as advertising advertisements, shopping advertisements, and electronic home appliance advertisements, and each category of advertisements can also be subdivided for each major. The category advertisement defines the advertisement category, and the segmented advertisement defines the advertisement category based on the corresponding large category advertisement. If the advertisement category of the travel advertisement is defined as T, the advertisement category of the travel agency advertisement subdivided under the travel advertisement is defined as T ₁ , the advertisement category of the hotel advertisement is defined as T ₂ , and the advertisement category of the travel city/scenic advertisement is defined as T ₃ The advertisement category of the shopping advertisement is defined as S, the advertisement category of the tobacco and alcohol advertisement segmented under the shopping advertisement is defined as S ₁ , the advertisement category of the food advertisement is defined as S ₂ , and the advertisement category of the appliance/electronic product is defined as S ₃ Etc., each of the sub-segments have a corresponding weight, and the ad category and its corresponding weight can be described as T(T ₁ , x ₁ ), (T ₂ , x ₂ ), (T ₃ , x ₃ S(S ₁ , x ₄ ), (S ₂ , x ₅ ), (S ₃ , x ₆ ). The user preference information is defined as P ₁ , P ₂ , P ₃ ... P _n , where n is determined according to the number of user preference information, and similarly, the user preference information and its corresponding weight can be described as P(P ₁ , y ₁ ), (P ₂ , y ₂ ), (P ₃ , y ₃ ) (P _n , y _n ). Based on the obtained x and y values, the cosine value calculation is performed on the advertisement category and the user preference information by using the calculation formula of the cosine similarity algorithm. When the calculated cosine value is closer to 1, the advertisement category and the user preference information are more Close, the higher the similarity.

S33: Determine whether the similarity is greater than a preset value.

The preset value is data preset by the system, and the preset value is a standard value for evaluating whether the similarity between the advertisement category of any advertisement and the user preference information reaches the associated advertisement. When the similarity between the advertisement category and the user preference information is greater than the preset value, the advertisement is determined to be closer to the user's preference, and the user is more likely to click the advertisement; when the advertisement category and the user preference information are not greater than the preset value, It is determined that the advertisement is not close to the user's preference, and may cause the user to decrease the click rate of the advertisement.

S34: If the similarity is greater than the preset value, determine that the advertisement is an associated advertisement.

In this embodiment, only the advertisement corresponding to the advertisement category whose similarity of the user preference information is greater than the preset value is used as the associated advertisement, so that the associated advertisement is closer to the user's preference, so as to facilitate the subsequent push of the associated advertisement to the user, It is easy to attract users' interest to increase the click rate of related ads.

S40: Push the associated advertisement to the client in real time, so that the client displays the associated advertisement in real time.

In this embodiment, the client sends an access request to the server, and when the control client displays the target webpage corresponding to the URL address in the access request, the server may implement the associated advertisement that is displayed by the display server, because the associated advertisement and the user The user preference information is associated, so that the associated advertisement is more likely to cause interest of the user, so that the user clicks to view the associated advertisement, thereby increasing the click rate of the advertisement.

Preferably, the client displaying the associated advertisement mode may be displayed in the APP in which the user account information is logged in or in the webpage when the user visits the webpage, or may be displayed on the terminal device corresponding to the terminal identifier carried in the access request, and the associated advertisement is displayed. The pop-up window is displayed so that the advertisement recommendation message does not affect the user's normal browsing of the webpage information. It can be understood that the associated advertisement is pushed to the client in real time, so that the client displays the associated advertisement in real time, and the associated advertisement can be viewed by the user who triggered the access request to improve the click rate of the advertisement; thereby avoiding triggering the access request. When the user is offline (ie, when leaving the client), the associated advertisement is pushed to other users, and the other users are not likely to click on the interest of the associated advertisement.

The advertisement real-time recommendation method can obtain the access request sent by the client in real time, identify the corresponding user preference information based on the request source of the access request, and then obtain the corresponding associated advertisement based on the user preference information, and push the associated advertisement in real time. The client that triggers the access request, so that the user can view the related advertisement in real time through the client. Since the associated advertisement is associated with the user preference information, and more in line with the user's interest, the click rate of the user clicking the associated advertisement can be improved. The purpose of pushing ads.

It should be understood that the size of the sequence of the steps in the above embodiments does not mean that the order of execution is performed. The order of execution of each process should be determined by its function and internal logic, and should not be construed as limiting the implementation process of the embodiments of the present application.

Example 2

FIG. 5 is a schematic block diagram showing an advertisement real-time recommendation device corresponding to the advertisement real-time recommendation method in the first embodiment. As shown in FIG. 5, the advertisement real-time recommendation device includes an access request acquisition module 10, a user preference information acquisition module 20, an associated advertisement acquisition module 30, and an associated advertisement recommendation module 40. The implementation functions of the access request acquisition module 10, the user preference information acquisition module 20, the associated advertisement acquisition module 30, and the associated advertisement recommendation module 40 correspond one-to-one with the steps corresponding to the advertisement real-time recommendation method in the embodiment. To avoid redundancy, the implementation The examples are not detailed one by one.

The access request obtaining module 10 is configured to obtain an access request sent by the client in real time, and the access request includes a request source identifier.

The user preference information obtaining module 20 is configured to acquire user preference information corresponding to the request source identifier based on the access request.

The associated advertisement obtaining module 30 is configured to acquire an associated advertisement corresponding to the user preference information based on the user preference information.

The associated advertisement recommendation module 40 is configured to push the associated advertisement to the client in real time, so that the client displays the associated advertisement in real time.

Preferably, the user preference information obtaining module 20 includes a user source identification determining unit 211, an existing user portrait data obtaining unit 212, an existing favorite information determining unit 213, a first user preference information acquiring unit 214, and a similar person. The group search unit 215, the second user preference information acquisition unit 216, the first webpage preference label acquisition unit 217, and the third user preference information acquisition unit 218; the user preference information acquisition module 20 further includes a terminal source identification determination unit 221 and a second The web page preference tag acquisition unit 222 and the fourth user preference information acquisition unit 223.

The user source identifier determining unit 211 is configured to determine, according to the access request, whether the request source identifier is a user identifier.

The existing user portrait data obtaining unit 212 is configured to query the existing user portrait data based on the user identifier if the request source identifier is the user identifier.

The favorite information judging unit 213 is configured to judge whether the existing user image data contains the existing preference information.

The first user preference information acquiring unit 214 is configured to use the existing favorite information as the user preference information if the existing user image data includes the existing favorite information.

The similarity group searching unit 215 is configured to search for similar people based on the existing user portrait data if the existing user image data does not contain the existing preference information.

The second user preference information obtaining unit 216 is configured to use the common preference information corresponding to the similar crowd as the user preference information.

The webpage preference tag first obtaining unit 217 is configured to search for the corresponding at least one historical webpage data based on the user identifier if the existing user portrait data does not include the existing favorite information, and each historical webpage data corresponds to a favorite tag.

The third user preference information obtaining unit 218 is configured to perform statistical analysis on the favorite tags corresponding to the historical webpage data, and obtain key preference tags to determine user preference information.

The terminal source identifier determining unit 221 is configured to determine, according to the access request, whether the request source identifier is a terminal identifier.

The second webpage preference tag obtaining unit 222 is configured to search for the corresponding at least one historical webpage data based on the terminal identifier if the source identifier is the terminal identifier, and each historical webpage data has a corresponding favorite label.

The fourth user preference information obtaining unit 223 is configured to perform statistical analysis on the favorite tags corresponding to the historical webpage data, and obtain key preference tags to determine user preference information.

Preferably, the related advertisement acquisition module 30 includes an advertisement category determination unit 31, an advertisement category similarity determination unit 32, a similarity determination unit 33, and an associated advertisement determination unit 34.

The advertisement category determining unit 31 is configured to perform keyword extraction on the advertisement to determine an advertisement category of the advertisement.

The advertisement category similarity determining unit 32 is configured to calculate the similarity between the advertisement category and the user preference information.

The similarity determining unit 33 is configured to determine whether the similarity is greater than a preset value.

The associated advertisement determining unit 34 is configured to determine that the advertisement is an associated advertisement if the similarity is greater than a preset value.

In the advertisement real-time recommendation device provided by the embodiment, the user preference information acquisition module 20 is configured to acquire user preference information corresponding to the request source identifier based on the access request.

If the source identifier carried by the access request is the user source identifier, the existing user portrait data is queried based on the user identifier, and it is determined whether the existing user portrait data contains the existing preference information. If the existing preference information is included, the existing preference information is used as the user preference information. If the existing user image data does not contain the existing preference information, the similar user is searched based on the existing user image data, and the common preference information corresponding to the similar group is used as the The user preference information; the corresponding at least one historical webpage data may be searched based on the user identifier, the favorite tag corresponding to each historical webpage data is determined, and all the favorite bookmarks are statistically analyzed, and the key preference tag is obtained to determine the user preference information.

If the source identifier carried by the access request is the terminal identifier, the at least one historical webpage data is searched according to the terminal identifier, and each historical webpage data has a corresponding favorite label, and all the favorite bookmarks are statistically analyzed to obtain a key favorite label. To determine user preferences.

The associated advertisement obtaining module 30 acquires the associated advertisement corresponding to the user preference information based on the obtained preference information, and determines that the advertisement is associated with the user when the determined advertisement category and the user's preference information similarity are greater than a preset value. Advertise and recommend to users. The related advertisements determined according to the user preference information are closer to the customer's needs, and when the advertisements are recommended to the corresponding users, the click rate of the recommended advertisements is increased.

Example 3

The embodiment provides a computer readable storage medium on which computer readable instructions are stored, and when the computer readable instructions are executed by the processor, the real-time recommendation method of the advertisement in Embodiment 1 is implemented. No longer. Alternatively, when the computer readable instructions are executed by the processor, the functions of the modules/units in the real-time recommendation device of the embodiment 2 are implemented. To avoid repetition, details are not described herein again.

Example 4

FIG. 6 is a schematic diagram of a terminal device according to an embodiment of the present application. As shown in FIG. 6, the terminal device 60 of this embodiment includes a processor 61, a memory 62, and computer readable instructions 63 stored in the memory 62 and operable on the processor 61, such as an advertisement real-time recommendation program. The processor 61 implements various steps of the advertisement real-time recommendation method in Embodiment 1 when the computer readable instructions 63 are executed, such as steps S10 to S40 shown in FIG. Alternatively, the processor 61 implements the functions of the modules/units in the advertisement real-time recommendation device in Embodiment 2 when the computer readable instructions 63 are executed.

Illustratively, computer readable instructions 63 may be partitioned into one or more modules/units, one or more modules/units being stored in memory 62 and executed by processor 61 to complete the application. The one or more modules/units may be a series of computer readable instruction segments capable of performing a particular function for describing the execution of computer readable instructions 63 in the terminal device 60. For example, the computer readable instructions 63 may be divided into an access request acquisition module 10, a user preference information acquisition module 20, an associated advertisement acquisition module 30, and an associated advertisement recommendation module 40. The specific functions of each module are as follows:

The terminal device 60 can be a computing device such as a desktop computer, a notebook, a palmtop computer, and a cloud server. The terminal device may include, but is not limited to, a processor 61, a memory 62. It will be understood by those skilled in the art that FIG. 6 is only an example of the terminal device 60, and does not constitute a limitation on the terminal device 60, and may include more or less components than those illustrated, or combine some components, or different components. For example, the terminal device may further include an input/output device, a network access device, a bus, and the like.

The processor 61 may be a central processing unit (CPU), or may be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc. The general purpose processor may be a microprocessor or the processor or any conventional processor or the like.

The memory 62 may be an internal storage unit of the terminal device 60, such as a hard disk or memory of the terminal device 60. The memory 62 may also be an external storage device of the terminal device 60, such as a plug-in hard disk provided on the terminal device 60, a smart memory card (SMC), a Secure Digital (SD) card, and a flash memory card (Flash). Card) and so on. Further, the memory 62 may also include both an internal storage unit of the terminal device 60 and an external storage device. The memory 62 is used to store computer readable instructions as well as other programs and data required by the terminal device. The memory 62 can also be used to temporarily store data that has been or will be output.

It will be apparent to those skilled in the art that, for convenience and brevity of description, only the division of each functional unit and module described above is exemplified. In practical applications, the above functions may be assigned to different functional units as needed. The module is completed, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or may be each Units exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

The integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, the present application implements all or part of the processes in the foregoing embodiments, and may also be implemented by computer readable instructions, which may be stored in a computer readable storage medium. The computer readable instructions, when executed by a processor, may implement the steps of the various method embodiments described above. Wherein, the computer readable instructions comprise computer readable instruction code, which may be in the form of source code, an object code form, an executable file or some intermediate form or the like. The computer readable medium can include any entity or device capable of carrying the computer readable instruction code, a recording medium, a USB flash drive, a removable hard drive, a magnetic disk, an optical disk, a computer memory, a read only memory (ROM, Read-Only) Memory), random access memory (RAM), electrical carrier signals, telecommunications signals, and software distribution media. It should be noted that the content contained in the computer readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in a jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, computer readable media It does not include electrical carrier signals and telecommunication signals.

The above-mentioned embodiments are only used to explain the technical solutions of the present application, and are not limited thereto; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still implement the foregoing embodiments. The technical solutions described in the examples are modified or equivalently replaced with some of the technical features; and the modifications or substitutions do not deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in Within the scope of protection of this application.

Claims

An advertisement real-time recommendation method, which is characterized by comprising:

Acquiring an access request sent by the client in real time, where the access request includes a request source identifier;

Acquiring user preference information corresponding to the request source identifier based on the access request;

Acquiring an associated advertisement corresponding to the user preference information based on the user preference information;

Pushing the associated advertisement to the client in real time, so that the client displays the associated advertisement in real time.
The advertisement real-time recommendation method according to claim 1, wherein the obtaining the user preference information corresponding to the request source identifier based on the access request comprises:

Determining, according to the access request, whether the request source identifier is a user identifier;

If the request source identifier is the user identifier, querying existing user portrait data based on the user identifier;

Determining whether the existing user portrait data contains existing preference information;

If the existing user portrait data includes the existing preference information, the existing preference information is used as the user preference information.
The advertisement real-time recommendation method according to claim 2, wherein the obtaining the user preference information corresponding to the request source identifier based on the access request further includes:

If the existing user portrait data does not include the existing preference information, searching for a similar crowd based on the existing user portrait data;

The common preference information corresponding to the similar crowd is used as the user preference information.
The advertisement real-time recommendation method according to claim 2, wherein the obtaining the user preference information corresponding to the request source identifier based on the access request further includes:

If the existing user profile data does not include the existing favorite information, searching for the corresponding at least one historical webpage data based on the user identifier, each of the historical webpage data corresponding to a favorite tag;

Perform statistical analysis on the favorite tag corresponding to the historical webpage data, and obtain a key preference tag to determine the user preference information.
The advertisement real-time recommendation method according to claim 1, wherein the obtaining the user preference information corresponding to the request source identifier based on the access request comprises:

Determining, according to the access request, whether the request source identifier is a terminal identifier;

And if the request source identifier is the terminal identifier, searching, according to the terminal identifier, the corresponding at least one historical webpage data, where each of the historical webpage data corresponds to a favorite label;

Perform statistical analysis on the favorite tag corresponding to the historical webpage data, and obtain a key preference tag to determine the user preference information.
The advertisement real-time recommendation method according to claim 1, wherein the access request further comprises a URL address;

The obtaining, according to the access request, the user preference information corresponding to the request source identifier, the method further includes: performing labeling processing on all webpages on the website, so that each of the webpages carries a favorite label;

The obtaining the user preference information corresponding to the request source identifier based on the access request, including: searching for a corresponding target webpage based on the URL address of the access request, and matching the favorite webpage of the target webpage As the user preference information.
The advertisement real-time recommendation method according to claim 1, wherein the acquiring the related advertisement corresponding to the user preference information based on the user preference information comprises:

Performing keyword extraction on the advertisement to determine an advertisement category of the advertisement;

Calculating a similarity between the advertisement category and the user preference information;

Determining whether the similarity is greater than the preset value;

If the similarity is greater than the preset value, determining that the advertisement is the associated advertisement.
An advertisement real-time recommendation device, comprising:

An access request obtaining module, configured to acquire an access request sent by the client in real time, where the access request includes a request source identifier;

a user preference information obtaining module, configured to acquire user preference information corresponding to the request source identifier based on the access request;

An associated advertisement obtaining module, configured to acquire an associated advertisement corresponding to the user preference information based on the user preference information;

The associated advertisement recommendation module is configured to push the associated advertisement to the client in real time, so that the client displays the associated advertisement in real time.
A terminal device comprising a memory, a processor, and computer readable instructions stored in the memory and operable on the processor, wherein the processor executes the computer readable instructions as follows step:

Acquiring an access request sent by the client in real time, where the access request includes a request source identifier;

Acquiring user preference information corresponding to the request source identifier based on the access request;

Acquiring an associated advertisement corresponding to the user preference information based on the user preference information;

Pushing the associated advertisement to the client in real time, so that the client displays the associated advertisement in real time.
The terminal device according to claim 9, wherein the obtaining the user preference information corresponding to the request source identifier based on the access request comprises:

Determining, according to the access request, whether the request source identifier is a user identifier;

If the request source identifier is the user identifier, querying existing user portrait data based on the user identifier;

Determining whether the existing user portrait data contains existing preference information;

If the existing user portrait data includes the existing preference information, the existing preference information is used as the user preference information;

If the existing user portrait data does not include the existing preference information, searching for a similar crowd based on the existing user portrait data;

The common preference information corresponding to the similar crowd is used as the user preference information.
The terminal device according to claim 10, wherein the obtaining the user preference information corresponding to the request source identifier based on the access request further includes:

If the existing user profile data does not include the existing favorite information, searching for the corresponding at least one historical webpage data based on the user identifier, each of the historical webpage data corresponding to a favorite tag;

Perform statistical analysis on the favorite tag corresponding to the historical webpage data, and obtain a key preference tag to determine the user preference information.
The terminal device according to claim 9, wherein the obtaining the user preference information corresponding to the request source identifier based on the access request comprises:

Determining, according to the access request, whether the request source identifier is a terminal identifier;

And if the request source identifier is the terminal identifier, searching, according to the terminal identifier, the corresponding at least one historical webpage data, where each of the historical webpage data corresponds to a favorite label;

Perform statistical analysis on the favorite tag corresponding to the historical webpage data, and obtain a key preference tag to determine the user preference information.
The terminal device according to claim 9, wherein the access request further comprises a URL address;

The obtaining, according to the access request, the user preference information corresponding to the request source identifier, the method further includes: performing labeling processing on all webpages on the website, so that each of the webpages carries a favorite label;

The obtaining the user preference information corresponding to the request source identifier based on the access request, including: searching for a corresponding target webpage based on the URL address of the access request, and matching the favorite webpage of the target webpage As the user preference information.
The terminal device according to claim 9, wherein the acquiring the associated advertisement corresponding to the user preference information based on the user preference information comprises:

Performing keyword extraction on the advertisement to determine an advertisement category of the advertisement;

Calculating a similarity between the advertisement category and the user preference information;

Determining whether the similarity is greater than the preset value;

If the similarity is greater than the preset value, determining that the advertisement is the associated advertisement.
A computer readable storage medium storing computer readable instructions, wherein the computer readable instructions, when executed by a processor, implement the following steps:

Acquiring an access request sent by the client in real time, where the access request includes a request source identifier;

Acquiring user preference information corresponding to the request source identifier based on the access request;

Acquiring an associated advertisement corresponding to the user preference information based on the user preference information;

Pushing the associated advertisement to the client in real time, so that the client displays the associated advertisement in real time.
The computer readable storage medium according to claim 15, wherein the obtaining the user preference information corresponding to the request source identifier based on the access request comprises:

Determining, according to the access request, whether the request source identifier is a user identifier;

If the request source identifier is the user identifier, querying existing user portrait data based on the user identifier;

Determining whether the existing user portrait data contains existing preference information;

If the existing user portrait data includes the existing preference information, the existing preference information is used as the user preference information;

If the existing user portrait data does not include the existing preference information, searching for a similar crowd based on the existing user portrait data;

The common preference information corresponding to the similar crowd is used as the user preference information.
The computer readable storage medium according to claim 16, wherein the obtaining the user preference information corresponding to the request source identifier based on the access request further comprises:

If the existing user profile data does not include the existing favorite information, searching for the corresponding at least one historical webpage data based on the user identifier, each of the historical webpage data corresponding to a favorite tag;

Perform statistical analysis on the favorite tag corresponding to the historical webpage data, and obtain a key preference tag to determine the user preference information.
The computer readable storage medium according to claim 15, wherein the obtaining the user preference information corresponding to the request source identifier based on the access request comprises:

Determining, according to the access request, whether the request source identifier is a terminal identifier;

And if the request source identifier is the terminal identifier, searching, according to the terminal identifier, the corresponding at least one historical webpage data, where each of the historical webpage data corresponds to a favorite label;

Perform statistical analysis on the favorite tag corresponding to the historical webpage data, and obtain a key preference tag to determine the user preference information.
The computer readable storage medium of claim 15 wherein the access request further comprises a URL address;

The obtaining, according to the access request, the user preference information corresponding to the request source identifier, the method further includes: performing labeling processing on all webpages on the website, so that each of the webpages carries a favorite label;

The obtaining the user preference information corresponding to the request source identifier based on the access request, including: searching for a corresponding target webpage based on the URL address of the access request, and matching the favorite webpage of the target webpage As the user preference information.
The computer readable storage medium according to claim 15, wherein the obtaining an associated advertisement corresponding to the user preference information based on the user preference information comprises:

Performing keyword extraction on the advertisement to determine an advertisement category of the advertisement;

Calculating a similarity between the advertisement category and the user preference information;

Determining whether the similarity is greater than the preset value;

If the similarity is greater than the preset value, determining that the advertisement is the associated advertisement.